Global kill-switch standard protocol system and method for ai robots
The global kill-switch protocol system addresses the limitations of existing AI robot shutdown systems by providing a standardized command frame, multi-path communication, and blockchain record-keeping to ensure secure, rapid, and transparent shutdowns across diverse AI robots.
Patent Information
- Application Number
- KR1020260053370
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2026-03-24
- Publication Date
- 2026-11-06
- Estimated Expiration
- 2046-03-24
AI Technical Summary
Existing emergency stop systems for AI robots are limited by manufacturer-specific implementation, lack of international standards, reliance on physical buttons or short-range communication, and insufficient legal evidence and safety protocols, making them difficult to control collectively and securely across different manufacturers and environments.
A global kill-switch standard protocol system that includes a standardized kill-switch command frame, multi-path communication network, multi-factor authentication, and blockchain-based record-keeping to ensure interoperability, rapid command propagation, and secure, authorized shutdowns across various AI robots.
Enables integrated, secure, and rapid shutdown of AI robots worldwide, minimizing secondary damage and ensuring legal transparency and accountability through standardized commands, multiple communication paths, and multi-stage authentication.
Smart Images

Figure 112026036003015-PAT00001_ABST
Abstract
Description
Technology Field
[0001] The present disclosure relates to a global kill-switch standard protocol system and method for AI robots. More specifically, the AI robots of the present disclosure are physical AIs that include AI robots having a physical entity and interacting directly with the real world, such as humanoid robots, embodied AI, and human AI robots. Background Technology
[0002] As the autonomy of Physical AI robots (such as humanoid robots, embodied AI, and human AI robots) that possess physical substance and interact directly with the real-world environment increases and their utilization in various fields grows, the possibility that these robots could escape human control and pose potential dangers is increasing. In particular, there are concerns about damage caused by robot malfunctions, hacking, or software defects in high-risk sectors such as military, autonomous driving, healthcare, and infrastructure management. To prevent these risks, countries are requiring emergency stop functions for robots, but existing emergency stop systems have several limitations.
[0003] First, existing emergency stop functions are implemented at the individual Physical AI robot or manufacturer level, making it difficult to collectively stop Physical AI robots from different manufacturers. This leads to a problem where stopping only a specific robot in an environment where robots from multiple manufacturers collaborate can cause secondary accidents.
[0004] Second, existing systems are implemented using physical buttons or short-range wireless communication, which limits remote emergency stopping capabilities. This makes emergency response difficult for robots operating in unmanned environments or internationally.
[0005] Third, the lack of internationally unified protocols makes it difficult for regulatory agencies in each country to enforce the blocking of physical AI robots from other nations. This creates a regulatory blind spot amidst the increasing trend of international operation of physical AI robots.
[0006] Fourth, emergency stop records are stored only within the robot, so they lack validity as evidence in legal disputes, and there are no means to prevent the abuse of emergency stop.
[0007] Fifth, additional risks may arise because safe restart procedures for Physical AI robots after an emergency shutdown are not standardized. To address this issue, there is a need to develop a standard kill-switch protocol that can be integrally applied at a global level to various Physical AI robots, such as humanoid robots and embodied AI. The problem to be solved
[0008] The purpose of one embodiment is to provide a global kill-switch standard communication protocol that can be integrally applied to various types of Physical AI robots (humanoid robots, embodied AI robots, human AI robots, etc.) differing in manufacturer, operating system, and communication method. The purpose of one embodiment is to provide an authority management system that classifies the authority to issue kill-switch commands into multiple stages and defines authentication procedures for each stage, thereby preventing the kill-switch from being activated without legitimate authority. The purpose of one embodiment is to ensure communication stability by establishing a multipath propagation network to propagate kill-switch commands at a global level in milliseconds. The purpose of one embodiment is to provide a safe stop protocol that minimizes secondary damage caused by sudden shutdown by applying a safe stop sequence according to the robot's current state when the kill-switch is executed. The purpose of one embodiment is to ensure transparency and legality by recording the reason for the kill-switch activation and the entire process on a blockchain. means of solving the problem
[0009] An AI robot global kill-switch standard protocol system according to one embodiment may include an electronic device and a Physical AI robot (including a Humanoid Robot, Embodied AI, or Human AI Robot, hereinafter referred to as an “AI Robot”) that communicates with the electronic device via a network, possesses a physical entity, and directly interacts with a real-world environment. The electronic device may generate a kill-switch command containing a kill-switch command frame in a standardized format that guarantees interoperability between machines independently of the manufacturer and operating system, based on the unique identification number of the AI Robot, and may generate and transmit a radio signal containing the generated kill-switch command to the AI Robot. The AI Robot may execute a forced blocking function of the AI Robot based on blocking level information included in the received kill-switch command frame.
[0010] In a system according to one embodiment, a kill-switch command frame may include an identification number of an AI robot, a blocking level, a blocking reason code, administrator identification information, and an expiration date, and an electronic device may generate transmission data encoded in a semantic data structure based on the kill-switch command frame.
[0011] In a system according to one embodiment, the electronic device can generate an authorization verification result with multiple authentications applied based on blocking level information, and provide an emergency bypass path that allows authentication of a single authorized person based on the input of a preset human threat condition.
[0012] In a system according to one embodiment, the network may include a global multipath propagation network including a plurality of independent communication paths, including satellite communication, terrestrial mobile communication, the Internet, and a dedicated emergency frequency band, and the electronic device may generate a multipath signal utilizing the plurality of independent communication paths simultaneously based on a kill-switch command.
[0013] In a system according to one embodiment, the AI robot can identify the current state information of the AI robot in real time based on its own sensor data acquired from a sensor mounted within the AI robot, and execute a safety stop sequence to minimize secondary damage caused by sudden blockage based on the identified current state information and blockage level information.
[0014] In a system according to one embodiment, the AI robot includes a hardware-based dedicated security chip that operates independently of the main operating system of the AI robot, and the kill-switch receiving and execution firmware within the dedicated security chip can independently execute a forced blocking function regardless of unauthorized firmware modification and the state of the operating system based on a received kill-switch command.
[0015] In a system according to one embodiment, an electronic device can generate a blockchain record transaction combined with a timestamp and an electronic signature based on the reason for activation of the kill-switch, administrator identity information, command propagation path, and execution result data of the AI robot.
[0016] In a system according to one embodiment, the AI robot generates a multi-stage safety verification result including hardware integrity self-diagnosis and software normal state verification based on the reception of a kill-switch release command, and can restore the operational authority of the AI robot in stages based on the multi-stage safety verification result being in a passed state. Effects of the invention
[0017] The AI robot global kill-switch standard protocol system according to the present invention can enable integrated kill-switching for robots using various manufacturers and operating systems through a kill-switch command frame of a standardized format that guarantees interoperability between machines independently of the manufacturer and operating system.
[0018] In addition, by utilizing multiple independent communication paths including satellite communication, terrestrial mobile communication, the Internet, and dedicated emergency frequency bands, kill-switch commands can be rapidly propagated at a global level, thereby minimizing response delays in emergency situations.
[0019] Furthermore, by preventing unauthorized activation of the kill switch through authorization verification using multi-factor authentication, while enabling rapid blocking via an emergency bypass route under conditions of threat to human life, it is possible to ensure both security and speed simultaneously.
[0020] In addition, by identifying the current status information of the AI robot in real time and executing a safety stop sequence to minimize secondary damage caused by sudden shutdown, the safe operation of the robot can be ensured.
[0021] Through blockchain-based transparency and audit modules, the legitimacy of kill switch activation can be transparently verified and legal evidentiary value secured, and safe restart can be guaranteed even after the kill switch is released through multi-stage safety verification.
[0022] These effects are in accordance with embodiments of the present invention, and the effects of the present invention are not limited to such literal descriptions but include everything that a person skilled in the art can infer from the present invention. Brief explanation of the drawing
[0023] FIG. 1 shows an example of a kill-switch standard protocol system according to one embodiment. FIG. 2 is a drawing for explaining the detailed operation of each module within an electronic device according to one embodiment. FIG. 3 is a diagram illustrating the detailed operation of each module within a Physical AI robot according to one embodiment. FIG. 4 shows an example of an electronic device according to one embodiment. Specific details for implementing the invention
[0024] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the actual form of implementation is not limited to the specific embodiments disclosed, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept described by the embodiments. Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, a first component may be named a second component, and similarly, a second component may be named a first component. When a component is referred to as being "connected" to another component, it should be understood that it may be directly connected to or joined to that other component, or that there may be other components in between. A singular expression includes a plural expression unless the context clearly indicates otherwise. In this specification, terms such as “comprising” or “having” are intended to specify the existence of the described features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this specification. Hereinafter, embodiments will be described in detail with reference to the accompanying drawings.In the description referring to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted.
[0026] FIG. 1 shows an example of a kill-switch standard protocol system according to one embodiment.
[0027] In at least one embodiment of the present disclosure, the AI robot global kill-switch standard protocol system may include an electronic device (110) and an AI robot (including a humanoid robot, embodied AI, physical AI, or human AI robot that has a physical entity and interacts directly with a real-world environment, hereinafter collectively referred to as “AI robot”) that communicates with the electronic device (110) via a network. The AI robot global kill-switch standard protocol system may utilize a network for communication between the electronic device (110) and the AI robot. The electronic device (110) may identify the unique identification number (ID) of the AI robot. The electronic device (110) may generate a kill-switch command based on the identified unique identification number (ID). The electronic device (110) may transmit the generated kill-switch command to the AI robot via the network. The AI robot may interpret the kill-switch command received via the network. The AI robot can execute a forced shutdown function according to an interpreted kill-switch command. The electronic device (110) and the AI robot may apply a standard protocol that ensures interoperability regardless of the manufacturer, operating system, or communication method. This configuration provides the effect of enabling integrated forced shutdown for Physical AI robots (humanoid robots, embodied artificial intelligence, human AI robots, etc.) of various manufacturers and operating systems.
[0028] The target robot can be identified by comparing the unique identification number (ID) with an internal database or an external authentication server. The electronic device (110) can refer to a standardized command frame format applicable regardless of the manufacturer and operating system. The electronic device (110) can generate a kill-switch command frame according to the standardized command frame format. The electronic device (110) may include the AI robot's identification number, block level, block reason code, administrator identification information, and validity period in the generated kill-switch command frame. The electronic device (110) can encode the generated kill-switch command frame into a semantic data structure. The electronic device (110) can convert the encoded command frame into a propagation signal. This series of processes provides the effect of generating integrated and consistent forced block commands for various Physical AI robots, such as Humanoid Robots, Embodied AI, and Human AI Robots, regardless of the manufacturer and operating system.
[0029] In at least one embodiment of the present disclosure, the electronic device (110) may generate an authorization verification result with multi-factor authentication applied based on block level information. The electronic device (110) may receive block level information. The electronic device (110) may identify the required authorization steps for each block level. The electronic device (110) may perform a multi-factor authentication procedure by combining multiple authentication means, such as biometric authentication, hardware tokens, and time-based one-time passwords (TOTP), for authorization verification. The electronic device (110) may generate an authorization verification result depending on the success or failure of the multi-factor authentication procedure. The electronic device (110) may include the authorization verification result in a kill-switch command frame or store it separately. This multi-factor authentication-based authorization verification provides the effect of preventing unauthorized activation and ensuring that only authorized personnel can issue commands for each block level.
[0030] In one embodiment, the electronic device (110) may provide an emergency bypass path that allows authentication by a single authorized person based on the input of a pre-set threat to life condition. The electronic device (110) may determine whether a threat to life condition has been input. If the threat to life condition is satisfied, the electronic device (110) may activate an emergency bypass path that allows the activation of a kill-switch command with only the authentication of a single authorized person instead of multiple authentications. When the emergency bypass path is activated, the electronic device (110) may simultaneously verify the biometric information and location information of the issuer. The electronic device (110) may require a post-execution regular authentication procedure for the command activated via the emergency bypass path. If regular authentication is not performed, the electronic device (110) may automatically release the kill-switch command and initiate an abuse of authority review. Such an emergency bypass path enables rapid blocking in emergency situations, such as threats to life, while providing the effect of preventing abuse through post-execution control.
[0031] In one embodiment, the electronic device (110) can generate transmission data encoded in a semantic data structure based on a kill-switch command frame. The electronic device (110) can convert each field of the generated kill-switch command frame into a semantic data structure, such as JSON-LD. The electronic device (110) can ensure interoperability between machines by generating transmission data encoded in a semantic data structure. The electronic device (110) can transmit the encoded transmission data to various Physical AI robot systems, such as humanoid robots, embodied AI robots, and human AI robots, via a network. Such encoding of the semantic data structure provides the effect of ensuring consistency in command interpretation between heterogeneous Physical AI robot systems.
[0032] In at least one embodiment of the present disclosure, the electronic device (110) can generate a blockchain record transaction combined with a timestamp and a digital signature based on the reason for activating the kill switch, administrator identity information, command propagation path, and execution result data of the AI robot. The electronic device (110) can collect the reason for activating the kill switch command, administrator identity information, command propagation path, and execution result data of the AI robot. The electronic device (110) can generate a blockchain record transaction by combining the collected information with a timestamp and a digital signature. The electronic device (110) can upload the generated blockchain record transaction to a blockchain network. The electronic device (110) can transparently track the entire process of activation, propagation, execution, and release of the kill switch command through the blockchain record. The generation of such a blockchain record transaction provides the effect of contributing to the verification of the legitimacy of the command and securing legal evidentiary capacity.
[0033] In one embodiment, the electronic device (110) may refer to a standardized format applicable regardless of the manufacturer and operating system. The electronic device (110) may store a command frame format defined by an international standardization organization (ISO / IEC, ITU-T, etc.). The electronic device (110) may determine the field structure and data type of the command frame based on the stored standardized format. If there are special fields specific to the manufacturer or operating system, the electronic device (110) may apply a mapping table to maintain compatibility with the standard fields. This reference to the standardized format provides the effect of enabling consistent command interpretation for Physical AI robots (humanoid robots, embodied artificial intelligence, human AI robots, etc.) of various manufacturers and operating systems.
[0034] In at least one embodiment of the present disclosure, an electronic device (110) can generate a kill-switch command frame according to a standardized format. The electronic device (110) can identify the required fields and optional fields of the standardized format. The electronic device (110) can complete the command frame by inputting or querying data corresponding to each field. The electronic device (110) can verify the data integrity of the generated command frame. The electronic device (110) can transmit the verified command frame to a subsequent encoding and propagation step. This generation of a command frame based on a standardized format provides the effect of increasing the interoperability and reliability of the command.
[0035] In at least one embodiment of the present disclosure, the kill-switch command frame may include an AI robot identification number, a blocking level, a blocking reason code, administrator identification information, and a validity period. The electronic device (110) may input the unique identification number (ID) of the target robot into the AI robot identification number field. The electronic device (110) may input a corresponding blocking level value from Level 1 to 5 into the blocking level field. The electronic device (110) may input a reason value selected from a predefined code table into the blocking reason code field. The electronic device (110) may input the identity and authentication information of the issuer into the administrator identification information field. The electronic device (110) may input the start and end times of the command application into the validity period field. The electronic device (110) may include additional parameters such as a geographical range, robot type, and ethics grade as needed. This configuration of fields provides the effect of clearly specifying the target, scope, authority, and validity of the command.
[0036] In at least one embodiment of the present disclosure, the electronic device (110) can generate an authorization verification result with multi-factor authentication applied based on blocking level information. The electronic device (110) can check the required authorization level for each blocking level in a predefined authorization table. The electronic device (110) can perform a multi-factor authentication procedure by combining multiple authentication means, such as biometric authentication, hardware tokens, and TOTP, for each level from 1 to 5. The electronic device (110) can generate an authorization verification result depending on the success or failure of the multi-factor authentication procedure. The electronic device (110) can include the authorization verification result in a command frame or store it separately. This multi-factor authorization verification provides the effect of ensuring that only authorized persons can issue commands for each blocking level.
[0037] In one embodiment, the electronic device (110) may provide an emergency bypass path that allows authentication by a single authorized person based on the input of a pre-set threat to life condition. The electronic device (110) may determine whether a threat to life condition has been input. If the threat to life condition is satisfied, the electronic device (110) may activate an emergency bypass path that allows the activation of a kill-switch command with only the authentication of a single authorized person instead of multiple authentications. When the emergency bypass path is activated, the electronic device (110) may simultaneously verify the biometric information and location information of the issuer. The electronic device (110) may require a post-execution regular authentication procedure for the command activated via the emergency bypass path. If regular authentication is not performed, the electronic device (110) may automatically release the kill-switch command and initiate an abuse of authority review. Such an emergency bypass path enables rapid blocking in emergency situations, such as threats to life, while providing the effect of preventing abuse through post-execution control.
[0038] In one embodiment, the electronic device (110) may receive input data to generate a kill-switch command frame. The electronic device (110) may receive input data from a user or an external system, such as a unique identification number (ID), block level, block reason, and administrator information. The electronic device (110) may map the received input data to a standardized command frame format. The electronic device (110) may verify the validity of the input data. The electronic device (110) may generate a command frame based on the verified input data. These input data reception and verification procedures provide the effect of increasing the accuracy and reliability of command generation.
[0039] In one embodiment, the electronic device (110) can receive a unique identification number (ID) of an AI robot. The electronic device (110) can receive the unique identification number (ID) through a network or a local input interface. The electronic device (110) can verify the format and whether there is a duplicate of the received unique identification number (ID). The electronic device (110) can input the verified unique identification number (ID) into the identification number field of a command frame. This reception and verification of the unique identification number provides the effect of clearly identifying the target robot of the command.
[0040] Integrity can be verified. Applying this standard format provides the effect of ensuring command consistency and reliability in various Physical AI robot environments, such as humanoid robots and embodied AI robots.
[0041] In at least one embodiment of the present disclosure, the electronic device (110) can generate transmission data encoded in a semantic data structure based on a kill-switch command frame. The electronic device (110) can convert each field of the command frame into a semantic data structure such as JSON-LD or RDF. The electronic device (110) can ensure interoperability between machines by generating transmission data encoded in a semantic data structure. The electronic device (110) can transmit the encoded transmission data to various robot systems via a network. The electronic device (110) can verify the validity of the semantic data structure. Such encoding of the semantic data structure provides the effect of ensuring consistency in command interpretation between heterogeneous systems.
[0042] In one embodiment, the electronic device (110) can convert a generated kill-switch command frame into a radio signal. The electronic device (110) may be equipped with a modulation circuit that converts the command frame into a digital signal. The electronic device (110) may convert the modulated digital signal into a frequency band or protocol suitable for each communication path. The electronic device (110) may transmit the converted radio signal to a transmission module. This radio signal conversion process provides the effect of enabling rapid propagation of commands in various communication environments.
[0043] In at least one embodiment of the present disclosure, an electronic device (110) can generate a kill-switch command as a radio signal through a plurality of independent communication paths, including satellite communication, terrestrial mobile communication, the Internet, and a dedicated emergency frequency band. The electronic device (110) can transmit the command signal to an LEO satellite mesh network through a satellite communication module. The electronic device (110) can transmit the command signal to a 5G / 6G network through a terrestrial mobile communication module. The electronic device (110) can transmit the command signal to a TCP / IP-based network through an Internet module. The electronic device (110) can transmit the command signal to a dedicated emergency frequency band, such as UHF / VHF, through an RF transmission module. The electronic device (110) can generate and transmit signals independently for each communication path. Such generation of multiple path signals provides the effect of increasing the certainty of command delivery even in the event of a single path failure.
[0044] In at least one embodiment of the present disclosure, the electronic device (110) can transmit a kill-switch command in parallel across multiple communication paths. The electronic device (110) can issue a simultaneous transmission command to multiple paths, such as satellite, mobile communication, the Internet, and emergency frequencies. The electronic device (110) can monitor the success or failure of transmission for each path in real time. If a transmission failure is detected in a specific path, the electronic device (110) can automatically activate an alternative node. The electronic device (110) can perform retransmission up to three times until transmission success is confirmed. This parallel transmission and retransmission logic provides the effect of maximizing the reliability of command propagation.
[0045] In one embodiment, the electronic device (110) can set a propagation priority so that kill-switch commands are transmitted preferentially even during network congestion. The electronic device (110) can set the propagation priority of the kill-switch command to be higher than general data communication through a network traffic management module. The electronic device (110) can apply a Quality of Service (QoS) policy to ensure that kill-switch commands are transmitted preferentially without delay even during network congestion. The electronic device (110) can monitor the results of the priority setting in real time. Such propagation priority setting provides the effect of ensuring the rapid arrival of commands in emergency situations.
[0046] In at least one embodiment of the present disclosure, the electronic device (110) may apply a triple anti-replay mechanism including a timestamp, a random number (Nonce), and a sequence number when propagating a kill-switch command. The electronic device (110) may include a timestamp indicating the current time in the command frame. The electronic device (110) may generate a unique random number (Nonce) for each command and include it in the command frame. The electronic device (110) may include a sequence number indicating the transmission order of the command in the command frame. The electronic device (110) may enable the receiving side to verify the timestamp, random number, and sequence number to detect repeated reception of the same command or replay attacks. Such a triple anti-replay mechanism provides the effect of effectively blocking tampering and repeated execution of the command.
[0047] In one embodiment, the electronic device (110) may apply end-to-end encryption to prevent tampering with commands during the propagation process. The electronic device (110) may encrypt the entire command frame using a public key-based encryption algorithm at the transmitting side. The electronic device (110) may decrypt the encrypted command using a corresponding secret key at the receiving side. The electronic device (110) may apply an encryption key management policy to prevent the encrypted command from being decrypted at an intermediate node. Such end-to-end encryption provides the effect of preventing tampering and eavesdropping during the propagation process of the command.
[0048] In at least one embodiment of the present disclosure, the electronic device (110) may perform a digital signature and integrity verification procedure to verify the integrity of a command. The electronic device (110) may include a digital signature generated with the sender's private key in the command frame. The electronic device (110) may verify the digital signature with the sender's public key at the receiving end. The electronic device (110) may perform integrity verification by calculating the hash value of the command frame. The electronic device (110) may block command execution and generate a warning if integrity verification fails. Such digital signature and integrity verification procedure provides the effect of ensuring the reliability and legitimacy of the command.
[0049] In one embodiment, the global multipath propagation network may include a satellite communication path. The global multipath propagation network may constitute a communication infrastructure connecting the ground and space through the satellite communication path. The satellite communication path may include various satellite networks, such as Low Earth Orbit (LEO) satellites, Medium Earth Orbit (MEO) satellites, and Geostationary Orbit (GEO) satellites. The satellite communication path may support bidirectional data transmission and reception between ground base stations and satellites. The satellite communication path may enable the propagation of kill-switch commands anywhere in the world without geographical restrictions. The satellite communication path may ensure the continuity of command propagation even in disaster situations where ground communication networks are disrupted.
[0050] In one embodiment, a global multipath propagation network can propagate kill-switch commands by utilizing satellite communication (LEO satellite mesh network). The global multipath propagation network can minimize the propagation delay of kill-switch commands by utilizing the low-latency characteristics of the LEO satellite mesh network. Since multiple satellites are interconnected in a mesh form, the LEO satellite mesh network can propagate commands via alternative paths even in the event of a failure of a specific satellite or ground station. The LEO satellite mesh network can independently verify the digital signature of the command at each satellite node. The LEO satellite mesh network provides global coverage, ensuring the delivery of kill-switch commands even in regions lacking ground infrastructure, such as polar regions, oceans, and deserts. The parallel operation of these satellite communication paths can provide the effect of maximizing the reliability and certainty of delivery of the global multipath propagation network.
[0051] In one embodiment, the global multipath propagation network may include a terrestrial mobile communication (5G / 6G) path. The global multipath propagation network can enable high-speed and low-latency communication in cities and densely populated areas through the terrestrial mobile communication path. The terrestrial mobile communication path may include base stations, repeaters, and core networks of a 5G or 6G mobile communication network. The terrestrial mobile communication path can enable continuous command reception through inter-cell handover even while the robot is moving. Since the terrestrial mobile communication path has a wide bandwidth and enables real-time transmission of large amounts of data, additional information such as status reports and authentication data can be transmitted in parallel in addition to kill-switch commands.
[0052] Path multiplexing and fault recovery are possible through base stations and relay nodes. 5G / 6G networks can support the rapid delivery of kill-switch commands in various environments, such as humanoid robots, embodied AI-based mobile robots, autonomous vehicles, and industrial physical AI robots. Utilizing these terrestrial mobile communication paths can provide the effect of enhancing emergency response capabilities in urban areas and industrial sites.
[0053] In one embodiment, the global multipath propagation network may include an Internet (TCP / IP) path. The global multipath propagation network may utilize wired and wireless network infrastructure through the Internet path. The Internet path may be interconnected with robots of various manufacturers and operating systems through a global network based on the TCP / IP protocol. The Internet path may support the transmission of kill-switch commands in various network environments, such as public networks, private networks, and VPNs. The Internet path may apply tunneling and port forwarding technologies to bypass network barriers such as firewalls, NAT, and proxies.
[0054] In one embodiment, a global multipath propagation network can propagate kill-switch commands by utilizing the Internet network. By leveraging the global coverage and scalability of the Internet network, the global multipath propagation network can enable simultaneous command propagation to multiple robots and control systems. The Internet network can enhance the speed and stability of command propagation by utilizing distributed infrastructure such as DNS, CDN, and cloud servers. The Internet network can prevent tampering and eavesdropping of commands through encrypted communication (SSL / TLS) and end-to-end authentication. The parallel operation of such Internet paths can provide the effect of maximizing command delivery rates in various network environments.
[0055] In one embodiment, the global multipath propagation network may include a dedicated emergency frequency (UHF / VHF) path. The global multipath propagation network may utilize dedicated emergency frequencies in the UHF (300–3000 MHz) and VHF (30–300 MHz) bands. The dedicated emergency frequency path may be preferentially activated in emergency situations such as disasters, war, or communication disruptions. The dedicated emergency frequency path is suitable for long-distance low-speed data transmission and has propagation characteristics that are resistant to terrain obstacles. The dedicated emergency frequency path may be interconnected with nationally allocated emergency communication networks, public safety networks, military communication networks, etc.
[0056] In one embodiment, a global multipath propagation network can propagate kill-switch commands by utilizing a dedicated emergency frequency band (UHF / VHF). Through the dedicated emergency frequency band, the global multipath propagation network can ensure the propagation of kill-switch commands even in situations where commercial communication networks are paralyzed. The dedicated emergency frequency band has minimal radio interference and congestion, enabling reliable transmission of commands. By linking with national disaster networks, emergency rescue networks, etc., the dedicated emergency frequency band can support rapid command transmission to control centers and field response teams. The introduction of such a dedicated emergency frequency path can provide the effect of maximizing communication survivability in extreme situations.
[0057] In at least one embodiment of the present disclosure, a global multipath propagation network can propagate a kill-switch command at a global level by simultaneously utilizing multiple independent communication paths. The global multipath propagation network can activate multiple paths in parallel, such as satellite communication, terrestrial mobile communication, the Internet, and dedicated emergency frequencies. Each communication path performs command propagation independently and can operate without mutual interference between paths. By operating multiple paths in parallel, even if a failure occurs in one path, the command can still be reached through another path. This parallel operation structure eliminates single-path dependency, thereby maximizing the reliability and availability of the entire system.
[0058] In at least one embodiment of the present disclosure, the global multipath propagation network may be operated complementarily so that each communication path operates independently, allowing commands to reach the target through another path even if a failure occurs in one path. Each communication path monitors the propagation status in real time and can automatically activate an alternative path in the event of a failure. The relay nodes of each path can independently verify the digital signature of the command to prevent forgery and replay attacks. Complementary operation can improve the probability of a kill-switch command reaching 99.999% or higher in response to the uncertainty of the global network environment. This structure can provide the effect of ensuring consistent command propagation in various countries and environments while maintaining compatibility with international standard communication protocols.
[0059] In one embodiment, the global multipath propagation network monitors the propagation status of each communication path in real time, automatically activates a replacement node for a specific path when a propagation failure is detected, and can perform up to three retransmissions until propagation success is confirmed. The global multipath propagation network can check the success of propagation for each path in real time. In a path where a propagation failure is detected, a pre-registered replacement node can be automatically activated. The replacement node can retransmit commands by bypassing the failure section. Retransmission can be repeated up to three times until propagation success is confirmed. During the retransmission process, a triple anti-replay mechanism, such as timestamps, random numbers, and sequence numbers, may be applied. This simultaneous propagation and retransmission structure can provide the effect of maximizing the reliability of command delivery in response to various risks such as network failures, congestion, and attacks.
[0060] In at least one embodiment of the present disclosure, a global multipath propagation network can simultaneously propagate a kill-switch command through multiple paths. Simultaneous propagation through multiple paths can reduce the time of arrival of the command and minimize overall system delay caused by delays or failures in specific paths. Simultaneous propagation can enable consistent command execution even in the event of duplicate reception by independently verifying the success of propagation in each path. The simultaneous propagation structure can provide the effect of ensuring rapid command execution in emergency situations by responding to the diversity and uncertainty of the global network environment.
[0061] In one embodiment, a global multipath propagation network can automatically request physical intervention from the nearest control center in the event of propagation failure in all paths. When propagation failure is detected in all paths, the global multipath propagation network can automatically transmit a signal requesting physical intervention to the control center. Based on the received request signal, the control center can immediately dispatch a field response team. The request for physical intervention may include the location of the target robot, identification number, and the reason for the propagation failure. The request for physical intervention is recorded on a blockchain along with a timestamp and can be used for post-audits and as legal evidence. This logic for requesting physical intervention can provide the effect of ensuring safe control of the robot even in extreme situations where network-based command propagation is impossible.
[0062] In one embodiment, a global multipath propagation network may be designed to enable a rapid physical response in the event of a propagation failure. Upon detection of a propagation failure, the control center may directly access the robot using personnel with on-site access or alternative means such as drones. This rapid physical response can prevent the spread of secondary damage and enable the safe stopping or retrieval of the robot. This design can provide the effect of securing a final safety control means even in situations where network-based control is completely impossible.
[0063] In one embodiment, the kill-switch command generation and standardization module (210) can map detailed commands, such as executable function restrictions, energy cutoff, communication cutoff, and physical locking, to a code table for each cutoff level. The code table can clearly define the sequence and parameters of execution commands corresponding to each cutoff level. The standardized code table can ensure consistent command interpretation even when new Physical AI robot models (humanoid robots, embodied AI robots, etc.) are added. The command set for each cutoff level can be subdivided according to the hardware and software characteristics of the Physical AI robot (walking mechanism of a humanoid robot, physical interaction function of an embodied AI, etc.). Code table-based management can provide the effect of minimizing the possibility of errors throughout the entire process of command generation, propagation, execution, and release.
[0064] In one embodiment, the international cooperation protocol module (250) defines the kill-switch authority relationship between the country of origin of the Physical AI robot and the country of registration when the Physical AI robot crosses a border and is operating in another country, wherein in the event of an immediate threat to human safety, the regulatory authority of the country of origin has the authority to prioritize the activation of the kill-switch up to Level 3 and notifies the country of registration afterward, and in the event of activation of the kill-switch at Level 4 or higher, joint approval by the regulatory authorities of the country of origin and the country of registration is required.
[0065] The international cooperation protocol module (250) can include information on the country of origin and country of registration of the robot within the command frame. The international cooperation protocol module (250) can automatically apply the priority activation authority of the regulatory body of the country of origin up to Level 3. When activation occurs at Level 4 or higher, the international cooperation protocol module (250) can activate a joint approval request and an approval waiting timer. If the joint approval waiting time is exceeded, the activation logic of the country of origin alone can be automatically executed. The international cooperation protocol module (250) can record the activation and approval history on the blockchain and use it as a basis for post-audit and dispute resolution.
[0066] In at least one embodiment of the present disclosure, the multi-level authority management and authentication module may classify the authority to issue kill-switch commands into five levels, from Level 1 to Level 5. The multi-level authority management and authentication module may predefine the blocking level that can be issued for each authority level. Level 1 authority may be granted to the owner or operator of a Physical AI robot (humanoid robot, embodied AI robot, etc.). Level 2 authority may be granted to a registration authority or a manufacturer's representative. Level 3 authority may be assigned to a national regulatory authority. Level 4 authority may be assigned to a national security agency or a designated high-level authority. Level 5 authority may be assigned to an international joint committee or a multinational approval body. The classification of authority levels may provide the effect of preventing unauthorized activation and abuse of authority.
[0067] In one embodiment, the multi-level authorization management and authentication module may apply multi-factor authentication combining biometric authentication, hardware tokens, and time-based one-time passwords at each level. The multi-level authorization management and authentication module may perform primary verification of the issuer's identity through biometric authentication (fingerprint, iris, facial recognition, etc.). Hardware tokens (USB keys, smart cards, etc.) may be used as a secondary authentication method. Time-based one-time passwords (TOTP) may be applied as a tertiary authentication method. Multi-factor authentication can independently verify the success of each authentication level. The multi-factor authentication system can minimize the risk of tampering with or theft of authentication data. The application of multi-factor authentication may be designed to strengthen the authentication procedure as the authorization level increases.
[0068] In one embodiment, the multi-level authorization management and authentication module may include an emergency bypass path that can be activated by emergency authentication of a single authorized person up to Level 3 in the event of an immediate threat to human life. The emergency bypass path enables the rapid activation of a blocking command when a threat situation occurs. The emergency bypass path may allow an authorized person at Level 3 or lower to activate a kill-switch command with only sole authentication. The emergency bypass path can prevent misuse by verifying the identity and location information of the issuer in real time. When the emergency bypass path is activated, a subsequent regular authentication procedure can be automatically triggered. The emergency bypass path can ensure transparency by recording the activation history on a blockchain.
[0069] In one embodiment, the multi-level authority management and authentication module (220) simultaneously verifies the biometric information and location information of the issuer during emergency bypass authentication, and the kill switch triggered by the emergency bypass authentication requires ratification by a regular authentication procedure within 24 hours of activation, but if ratification is not made, the kill switch is automatically released and an examination of abuse of authority against the issuer can be initiated.
[0070] The multi-level authority management and authentication module (220) can simultaneously collect and verify the issuer's biometric information (fingerprint, iris, etc.) and location information (GPS, network location, etc.) during emergency bypass authentication. The kill-switch command triggered by emergency bypass authentication can be recorded on the blockchain along with an emergency trigger flag. The regular authentication procedure can send an automatic notification and authentication request to the issuer within 24 hours of triggering. If the regular authentication approval is not performed, the system can automatically release the kill-switch command and initiate an authority abuse review process against the issuer. The authority abuse review can be performed by comprehensively analyzing blockchain records, authentication logs, location data, etc.
[0071] In one embodiment, the AI robot can determine the current state information of the AI robot in real time based on its own sensor data obtained from a sensor mounted within the AI robot.
[0072] The AI robot can periodically collect data from built-in sensors (accelerometer, gyroscope, position sensor, environment sensor, etc.). The AI robot can convert the collected sensor data into reliable state information through noise filtering and preprocessing. The AI robot can classify various states in real time, such as movement, work, charging, and standby. The AI robot can ensure the timely execution of a safety stop sequence through real-time updates of state information. In particular, if the robot (120) is a humanoid robot, it is possible to determine the state more precisely by additionally acquiring physical sensor data unique to the human body structure, such as bipedal walking patterns, joint angles, finger positions, balance state, and center of gravity movement.
[0073] In one embodiment, the AI robot can execute a safety stop sequence to minimize secondary damage caused by sudden blockage based on identified current status information and blockage level information. The AI robot can execute a customized safety stop sequence (speed reduction, movement to a safe zone, tool withdrawal, etc.) step-by-step according to the status information and blockage level. The AI robot can monitor each step of the stop sequence in real time by linking it with sensor data. The AI robot can automatically report whether the stop sequence has been completed to internal logic and an external control system. The application of the safety stop sequence can provide the effect of significantly reducing the possibility of secondary damage, such as falling, collision, or interruption of the procedure.
[0074] In at least one embodiment of the present disclosure, a blockchain-based transparency and audit module may record the cause of activation of the kill switch, issuer identity information, authentication records, command propagation path, arrival time for each node, execution results of the target robot, and release records on a blockchain network along with timestamps and digital signatures. The blockchain-based transparency and audit module may generate the creation time of each kill switch command, issuer authentication information, command propagation path, and reception time of each relay node as a blockchain transaction. The blockchain record may include the command execution results (time of suspension completion, suspension status, release approval, etc.). The blockchain record may secure tamper-proofing and legal evidentiary capabilities by combining digital signatures and timestamps. Due to the distributed ledger structure, the blockchain-based record can guarantee data permanence and reliability even in the event of a single node failure.
[0075] In at least one embodiment of the present disclosure, a blockchain-based transparency and audit module may provide an audit trail capable of retrospectively verifying the legitimacy of a kill-switch command. For the audit trail, the blockchain-based transparency and audit module may automatically aggregate the reasons for issuing the command, evidence of threat situations, issuer authentication logs, robot status data, etc. The audit trail may support independent audit bodies or relevant authorities in viewing blockchain records in real time. The audit trail function enables retrospective verification of various issues, such as the issuance of unjust commands, abuse of authority, and system malfunctions. The audit trail may be utilized as an objective basis for legal judgment in the event of a dispute.
[0076] In at least one embodiment of the present disclosure, the blockchain-based transparency and audit module may include an objection and compensation procedure regarding an unjustified kill-switch activation. Upon receipt of an objection, the blockchain-based transparency and audit module may automatically collect all relevant data recorded on the blockchain and submit it to an audit body. If the activation is determined to be unjustified, relief procedures, such as restricting the authority of the issuer, compensating the owner of the damaged robot, and implementing measures to prevent recurrence, may be executed in stages. The compensation procedure may objectively calculate the scale of damage and the allocation of liability based on the blockchain records. The objection and compensation procedure may provide the effect of enhancing the reliability and transparency of the system.
[0077] In at least one embodiment of the present disclosure, the AI robot may include a hardware-based dedicated security chip that operates independently of the AI robot's main operating system.
[0078] AI robots can implement a secure environment physically separated from the main operating system (OS) by embedding a dedicated security chip (Secure Element) in hardware. The dedicated security chip may include an independent communication receiver, a command verification unit, and a driver control interface. The dedicated security chip can be designed to be physically embedded during manufacturing so that it cannot be removed afterwards. The dedicated security chip can prevent unauthorized firmware modification by applying hardware-based Secure Boot.
[0079] In at least one embodiment of the present disclosure, the kill-switch receiving and executing firmware within a dedicated security chip can independently execute a forced blocking function based on a received kill-switch command, regardless of unauthorized firmware modification or the state of the operating system. The kill-switch receiving and executing firmware can independently perform quadruple verification, including the digital signature of the command message, the issuer's authority, the command validity period, and the geographical scope. The firmware can receive and execute commands even when the operating system is hacked or disabled. The firmware can automatically trigger the activation of the kill-switch upon detecting an attempt at an unauthorized update. The independent execution structure of the firmware can provide the effect of maximizing the security and reliability of the entire system.
[0080] In one embodiment, the AI robot can generate multi-stage safety verification results, including hardware integrity self-diagnosis and software normal state verification, based on the reception of a kill-switch release command.
[0081] Upon receiving a kill-switch release command, the AI robot can perform a hardware integrity self-diagnosis in Step 1. In Step 2, the AI robot can verify the normal status of the software and AI models. In Step 3, the AI robot can perform sensor calibration and environmental perception function verification. In Step 4, the AI robot can conduct test runs within a limited range. The verification results of each step may be automatically reported to the registration authority.
[0082] In one embodiment, the AI robot may gradually restore operational authority based on the fact that the results of the multi-stage safety verification have passed. The AI robot may allow full resumption of operation only when all safety verification stages have been passed. The AI robot may gradually restore operational authority in the order of limited function restoration, test operation, and full resumption of operation. This phased restoration of authority can provide the effect of preventing additional risks caused by improper restart.
[0083] FIG. 2 is a drawing for explaining the detailed operation of each module within an electronic device (110) according to one embodiment.
[0084] In at least one embodiment of the present disclosure, the kill-switch command frame may be composed of a standardized field containing a unique identification number of the AI robot. The unique identification number field of the kill-switch command frame may include a unique identification value assigned to each robot in the data field. The unique identification number may be generated by combining multiple identification elements, such as a manufacturer, model, production serial number, or country registration number. The unique identification number field may be defined in an international standard identifier format, such as hexadecimal, decimal, or UUID. The unique identification number field may be allocated a fixed byte length within the command frame to ensure consistency in data parsing. The unique identification number field may be designed so that the command-receiving robot can quickly determine that it is the target. The unique identification number field may be linked with blockchain audit records, command execution logs, etc., and utilized as a reference value for post-tracing and verification. Standardization of the unique identification number field can prevent errors in command interpretation in diverse Physical AI robot environments, such as humanoid robots and embodied AI robots, which differ in manufacturers, operating systems, and communication methods.
[0085] In one embodiment, the unique identification number field of the kill-switch command frame can uniquely identify the target AI robot. The unique identification number field can prevent duplicate identification errors among multiple robots within the same network. The unique identification number field can be directly compared and verified by the target robot's hardware security chip or authentication module at the command frame reception stage. The unique identification number field can be utilized as a key reference value for preventing tampering and unauthorized execution of the command frame. The unique identification number field can be designed with compatibility with international standard (ISO / IEC, ITU-T) identification schemes in mind.
[0086] In one embodiment, a kill-switch command frame may be composed of a standardized field containing blocking level information. The blocking level information field may be defined as a separate data field within the command frame. The blocking level information may be expressed as step-by-step code values from Level 1 to Level 5. The blocking level information field may map detailed commands, such as function restrictions, energy cutoff, communication cutoff, and physical locking, which are executable for each level, to a code table. The blocking level information field may be designed to allow the robot's security chip or firmware to automatically select an execution sequence appropriate for the corresponding level upon command execution. The blocking level information field may provide flexibility for rapid blocking or step-by-step control in emergency situations. The blocking level information field may record the interpretation and execution results of the command frame step-by-step in a blockchain audit record.
[0087] In one embodiment, the cutoff level information field can specify step-by-step cutoff commands from Level 1 to Level 5. Level 1 can specify partial restriction of specific functions. Level 2 can specify the deactivation of major autonomous functions, such as autonomous driving and autonomous decision-making, and the switching to manual control. Level 3 can specify the cessation of all driving functions and the switching to safety mode. Level 4 can specify the cutoff of energy supply and external communication. Level 5 can specify the locking of physical drive units, activation of self-destruction prevention mode, and transmission of position fixation signals. Cutoff commands for each level can be consistently managed by a standardized code table.
[0088] In one embodiment, the kill-switch command generation and standardization module (210) defines the blocking levels of the kill-switch command as Level 1 (restriction of some specific functions), Level 2 (disabling of major autonomous functions such as autonomous driving and autonomous judgment and switching to manual control), Level 3 (stopping of all driving functions and switching to safety mode), Level 4 (cutting off energy supply and cutting off external communication), and Level 5 (locking of physical drive parts, activation of self-destruction prevention mode and transmission of position fixing signal), and can manage a specific set of execution commands corresponding to each blocking level in a standardized code table. The kill-switch command generation and standardization module (210) can clearly define detailed command sequences and parameters executable for each blocking level in the code table. The code table may include the order, conditions, and exception handling logic of execution commands corresponding to each blocking level. The code table may be designed with compatibility with international standard data models in mind. Command execution based on the code table can ensure an automated execution flow in the robot hardware and firmware. The code table can be flexibly expanded when adding new blocking levels or changing policies. Standardization of code tables can minimize command interpretation errors for Physical AI robots (humanoid robots, embodied AI, human AI robots, etc.) across countries and manufacturers.
[0089] In at least one embodiment of the present disclosure, a kill-switch command frame may be composed of a standardized field including a blocking reason code. The blocking reason code field may be defined as a separate data field within the command frame. The blocking reason code field can clearly distinguish various reasons, such as predefined threat types, malfunctions, regulatory violations, hacking detection, and emergencies, by code values. The blocking reason code field may be standardized with a unique numeric or character code for each reason. The blocking reason code field can support the rapid identification of the cause of the blocking by robots and control systems during command execution. The blocking reason code field can be utilized as evidence in blockchain audit records, post-hoc analysis, and legal disputes. The blocking reason code field may be designed with compatibility with international standard threat classification schemes in mind.
[0090] In one embodiment, the blocking reason code field can clearly distinguish various reasons, such as predefined threat types, malfunctions, and regulatory violations, by code values. For example, the blocking reason code field can be defined as '01' for hacking detection, '02' for human life threat, '03' for regulatory violation, '04' for system malfunction, '05' for emergency testing, etc. The blocking reason code field can be automatically included in the command frame interpretation and execution result report. If multiple reasons overlap, the blocking reason code field can include multiple code values in an array format.
[0091] In one embodiment, the kill-switch command frame may be composed of a standardized field containing administrator identification information. The administrator identification information field may be defined as a separate data field within the command frame. The administrator identification information field may include multiple authentication data, such as the identity, organization, authority level, and authentication token of the command issuer. The administrator identification information field may include data from various authentication methods, such as digital signatures, public keys, and biometric authentication data. The administrator identification information field may be designed to automatically perform authority verification of the issuer by the command receiving robot or control system. The administrator identification information field may be utilized as a reference value for identity verification in blockchain audit records, post-tracement, and authority abuse screening. The administrator identification information field may be designed with compatibility with international standard authentication schemes in mind.
[0092] In at least one embodiment of the present disclosure, the administrator identification information field may include authentication data capable of verifying the identity and authority level of the command issuer. The administrator identification information field may include, for example, an issuer ID, an agency code, an authority level, an electronic signature hash value, etc. The administrator identification information field may be utilized as a key reference value for preventing tampering and unauthorized activation of the command frame. The administrator identification information field may simultaneously include the issuer's biometric information and location information during emergency bypass authentication.
[0094] In at least one embodiment of the present disclosure, the multi-level authority management and authentication module (220) may classify the authority to issue kill-switch commands into five levels, from Level 1 (partial restriction of function) to Level 5 (physical locking), according to the level of blocking. The multi-level authority management and authentication module (220) may apply different identity of the authorized person, authentication method, and set of execution commands for each blocking level. Each level from Level 1 to Level 5 may be specifically defined according to the scope and intensity of the blocking. Level 1 may be subdivided into partial restriction of software function, Level 2 into deactivation of major autonomous functions, Level 3 into cessation of all operating functions, Level 4 into energy and communication blocking, and Level 5 into physical locking and position fixing. Each authority level may be consistently managed by a standardized code table. The scope of commands that can be issued and the execution conditions may be clearly distinguished for each authority level. The classification of authority levels may have the effect of ensuring a balance between preventing unauthorized activation and responding to emergency situations.
[0095] In at least one embodiment of the present disclosure, Level 1 may be issued by a robot owner or operator. A Level 1 authorized person may verify their identity through the robot's registration information and owner authentication. A Level 1 authorized person may issue only commands to restrict some functions. A Level 1 authorized person may prevent abuse of authority through a multi-factor authentication procedure.
[0096] In one embodiment, Level 2 may be authorized by a registration authority or a manufacturer. A Level 2 authorized person may verify their identity through certification by the manufacturer of the robot or an official registration authority. A Level 2 authorized person may issue commands to disable key autonomous functions and switch to manual control. A Level 2 authorized person may require a more stringent authentication procedure than Level 1.
[0097] In one embodiment, Level 3 may be issued by a national regulatory authority. A Level 3 authorized person may verify their identity through official certification by a national regulatory authority. A Level 3 authorized person may issue commands to stop all operating functions and switch to safe mode. Authorization and certification procedures for Level 3 authorized persons may be defined in accordance with national laws.
[0098] In one embodiment, Level 4 may be issued by a national security agency. A Level 4 authorized person may verify their identity through the national security agency's unique authentication system. A Level 4 authorized person may issue orders to cut off energy supply and cut off external communications. A Level 4 authorized person may have priority activation authority for rapid cutoff in emergency situations.
[0099] In one embodiment, Level 5 may be defined as a stage requiring approval from an international joint committee. A Level 5 authorized person may be granted authority through a multi-country approval procedure by an international joint committee. A Level 5 authorized person may issue the highest level of blocking commands, such as locking physical actuators, activating self-destruction prevention mode, and transmitting location-fixing signals. The activation of a Level 5 authorized person may be permitted only in situations of international conflict or transnational threat.
[0100] In one embodiment, each authority level can be managed in conjunction with a specific set of execution commands for each blocking level. The set of execution commands for each authority level may include standardized command codes and execution conditions. The set of execution commands can be finely adjusted according to the type of robot, the operating environment, and national regulations. The set of execution commands can be inserted as code values within a command frame to ensure interoperability between machines. The linkage of command sets by authority level can have a defensive effect against security threats such as malfunctions, hacking, and abuse of authority.
[0101] In one embodiment, the multi-level authorization management and authentication module (220) may apply multi-factor authentication combining biometric authentication, hardware tokens, and time-based one-time passwords at each level. The multi-level authorization management and authentication module (220) may require at least two types of authentication methods in parallel or sequentially at each authorization level. Biometric authentication, hardware tokens, and TOTPs can be verified independently of each other to enhance authentication reliability. The application of multi-factor authentication can effectively block abuse of authority and hacking attempts. The success or failure of each authentication method can be recorded and verified at each level prior to the issuance of a command.
[0102] In at least one embodiment of the present disclosure, biometric authentication may utilize various biometric information such as fingerprints, irises, and faces. Biometric authentication may generate a digital template of unique biometric characteristics to perform real-time comparative verification. Biometric authentication data may prevent the leakage of personal information through encrypted storage and transmission. Biometric authentication may have the effect of preventing identity forgery and proxy authentication of authorized persons.
[0103] In at least one embodiment of the present disclosure, the hardware token may include physical authentication means such as a FIDO authentication device or a smart card. The hardware token may transmit authentication data through various interfaces such as USB, NFC, and BLE. The hardware token may embed a unique identification value and an encryption chip to prevent loss, duplication, or theft. Hardware token authentication may have a defensive effect against network-based attacks.
[0104] In one embodiment, a time-based one-time password (TOTP) may use a one-time authentication value generated according to a standardized algorithm. The TOTP may generate a new authentication value every 30 or 60 seconds according to international standard algorithms such as RFC 6238. The TOTP may verify the validity of the authentication value based on time synchronization between the server and the client. The TOTP may prevent reuse attacks due to the short-term validity of the authentication value.
[0105] In one embodiment, a multi-factor authentication system can simultaneously prevent abuse of authority and enhance authentication reliability. The multi-factor authentication system can reject the entire authentication in the event of a single authentication failure by utilizing the complementary characteristics of each authentication method. The multi-factor authentication system can automatically store authentication history and failure records in a blockchain or audit system. The multi-factor authentication system can provide a defensive effect against various security threats, such as identity forgery of authorized persons, theft of authentication information, and insider attacks.
[0106] In at least one embodiment of the present disclosure, the multi-level authority management and authentication module (220) may include an emergency bypass path that can be activated by emergency authentication of a single authority up to a predetermined level in the event of an immediate threat to human life. The emergency bypass path may be designed to enable the activation of a kill-switch command by rapid authentication of a single authority at Level 3 or lower. The emergency bypass path may be automatically activated through a pre-registered list of authority holders and emergency situation determination logic. The emergency bypass path can simultaneously achieve rapid responsiveness and ensure safety of the system.
[0107] In at least one embodiment of the present disclosure, the issuer's biometric information and location information can be verified simultaneously during emergency bypass authentication. Emergency bypass authentication can be verified by merging the issuer's real-time location coordinates with biometric authentication data. Location information can utilize various location measurement technologies, such as GPS, Wi-Fi, and cellular networks. Simultaneous verification of biometric information and location information can prevent proxy authentication and remote forgery attempts by authorized persons.
[0108] In at least one embodiment of the present disclosure, a kill switch triggered by emergency bypass authentication may require ratification by a regular authentication procedure within 24 hours of triggering. When emergency bypass authentication is triggered, the system may automatically send a notification of the ratification request to a regular authorized person or a higher authority. The ratification request may be recorded with a timestamp on a blockchain or audit system. If ratification is not completed within 24 hours, an automatic release logic may be executed.
[0109] In one embodiment, if ratification is not granted, the kill switch may be automatically released, and an investigation into abuse of authority by the issuer may be initiated. The automatic release logic monitors the execution status of the kill switch command in real time and can immediately lift the block if ratification is not granted. The investigation into abuse of authority may comprehensively analyze the issuer's authentication history, location information, and evidence of emergency situations. The results of the investigation into abuse of authority may be linked to follow-up procedures, such as authority restriction, preventive training, and legal action.
[0110] In at least one embodiment of the present disclosure, an emergency detour route and a post-validation procedure can simultaneously ensure the speed and accountability of the system. The emergency detour route can have the effect of minimizing response delays in life-threatening situations. The post-validation procedure enables the tracking of responsibility and post-verification regarding abuse of authority and false orders. The simultaneous securing of speed and accountability can enhance the reliability and legal legitimacy of the system.
[0111] In one embodiment, the multi-level authorization management and authentication module (220) can perform authorization verification logic based on blocking level information. The authorization verification logic can check the level of the authorized person, whether authentication is successful, the command validity period, the geographical range, etc., step by step based on the blocking level code value within the command frame. The authorization verification logic can automatically match the authentication method and execution conditions required for each step. The authorization verification logic can determine in real time whether execution is possible for each blocking level.
[0112] In at least one embodiment of the present disclosure, the authority verification logic can comprehensively verify the issuer's authority level, the success of multi-factor authentication, the validity period of the command, and the geographical range. The authority verification logic can verify the authority to issue the command by querying the issuer's authority level from a database. The success of multi-factor authentication can determine final approval by integrating the result values of each authentication method. The validity period of the command can be verified for expiration by comparing timestamps. The geographical range can determine applicability by comparing location information within the command frame with the robot's current location.
[0113] In one embodiment, the authorization verification result can be used as a criterion value to determine whether execution is possible for each blocking level. The authorization verification result can be transmitted to a robot or network node prior to command execution to determine whether to approve or deny execution. The authorization verification result is automatically stored in the blockchain audit record and can be used as a basis for post-verification and dispute resolution. The authorization verification result can serve as a key criterion value for preventing unauthorized activation of the system and enhancing security.
[0114] In one embodiment, the multi-level authorization management and authentication module (220) can immediately execute a command containing an emergency code indicating an immediate threat to human life solely through digital signature verification. A command containing an emergency code may be indicated by a separate flag or code value within the command frame. When an emergency code is detected, the authorization verification logic may skip additional authentication procedures and determine whether to approve execution solely through digital signature verification. Digital signature verification can function as a core security procedure to prevent tampering with and unauthorized activation of the command.
[0115] In at least one embodiment of the present disclosure, if an emergency code is included, an additional authorization verification procedure can be omitted and a kill-switch command can be executed quickly. The rapid execution process can generate an execution approval signal within a few milliseconds after receiving the command. The rapid execution process can minimize response delays caused by external factors such as network congestion and authentication delays. The rapid execution process can have the effect of preventing casualties and secondary accidents in emergency situations.
[0116] In at least one embodiment of the present disclosure, rapid execution processing may have the effect of minimizing response delays in threat situations. Rapid execution processing may simultaneously enhance the real-time capability, safety, and reliability of the system. Rapid execution processing may be linked to post-event audit and accountability procedures after the emergency situation has ended.
[0117] In at least one embodiment of the present disclosure, the blockchain-based transparency and audit module (240) may record the reason for the activation of the kill switch on a blockchain network. The blockchain-based transparency and audit module (240) may record the activation reason code by including it in a standardized data field at the time the kill switch command is activated. Along with the activation reason code, the blockchain-based transparency and audit module (240) may record a description of the threat situation at the time of the activation request, the hash value of related evidence, and the identity information of the issuer. The blockchain-based transparency and audit module (240) may store the recorded activation reason in a way that prevents tampering through the distributed ledger structure of the blockchain network. The blockchain-based transparency and audit module (240) may clarify the legitimacy of the kill switch command and the subsequent liability through the recording of the activation reason.
[0118] In one embodiment, the blockchain-based transparency and audit module (240) can automatically collect the reason for activation, threat situation evidence, issuer authentication records, and target robot status data recorded on the blockchain and submit them to an independent review body when an objection is raised regarding the legitimacy of the kill-switch activation. The blockchain-based transparency and audit module (240) can automatically extract all relevant records stored on the blockchain upon receipt of an objection. The blockchain-based transparency and audit module (240) can provide the original or hash value of the threat situation evidence so that the independent review body can verify the authenticity of the data. The blockchain-based transparency and audit module (240) can objectively reconstruct the situation at the time of activation by linking the issuer authentication records and the target robot status data. The blockchain-based transparency and audit module (240) can ensure the speed and transparency of the objection procedure through an automatic submission function.
[0119] In one embodiment, the blockchain-based transparency and audit module (240) can perform relief procedures including restricting the authority of the issuer, calculating compensation for the victimized robot owner, and taking measures to prevent recurrence in the event that the issuance is determined to be improper. The blockchain-based transparency and audit module (240) can automatically receive the results of the judgment of an independent review body. When the issuance is determined to be improper, the blockchain-based transparency and audit module (240) can immediately restrict or temporarily suspend the authority level of the issuer. The blockchain-based transparency and audit module (240) can calculate the details of the victimized robot owner's losses based on blockchain records. As measures to prevent recurrence, the blockchain-based transparency and audit module (240) can automatically execute measures such as strengthening additional certification procedures, requiring completion of training, and providing system warning notifications. The blockchain-based transparency and audit module (240) can enable post-audits by additionally recording all details of the implementation of the relief procedures on the blockchain.
[0120] In one embodiment, the blockchain-based transparency and audit module (240) can provide an audit trail including the reason for triggering the kill switch. The blockchain-based transparency and audit module (240) can record the entire process of the kill switch command, including creation, authentication, propagation, execution, and release, in chronological order. The blockchain-based transparency and audit module (240) can ensure the continuity and reliability of the audit trail by assigning a unique identifier and timestamp to the record of each step. The blockchain-based transparency and audit module (240) can provide objective evidence to external auditing bodies or in legal disputes through the audit trail data.
[0121] In one embodiment, the blockchain-based transparency and audit module (240) can record the issuer's identity information and authentication records on a blockchain network. The blockchain-based transparency and audit module (240) can record the issuer's unique identification number, affiliated organization, authority level, and authentication method (biometric, token, TOTP, etc.) in standardized fields. The blockchain-based transparency and audit module (240) can include the authentication time, whether the authentication was successful or failed, and authentication device information in the authentication records. Since the identity and authentication information is stored in a blockchain distributed ledger, the blockchain-based transparency and audit module (240) can manage it in a state where tampering and deletion are impossible. The blockchain-based transparency and audit module (240) can clarify the legitimacy of the issuing authority and the locus of responsibility through the recording of identity and authentication information.
[0122] In at least one embodiment of the present disclosure, the blockchain-based transparency and audit module (240) may include an issuer authentication record and submit it to an independent review body upon objection. The blockchain-based transparency and audit module (240) may automatically extract the issuer authentication record and transmit it to an independent review body upon receipt of an objection. The blockchain-based transparency and audit module (240) may provide data along with an electronic signature and a timestamp so that the review body can verify the authenticity of the authentication record. The blockchain-based transparency and audit module (240) may automate the process of submitting the authentication record, thereby ensuring the speed and objectivity of the objection procedure.
[0123] In one embodiment, the blockchain-based transparency and audit module (240) can verify the legitimacy of a kill-switch command retrospectively through administrator identity and authentication information. The blockchain-based transparency and audit module (240) can verify the legality of issuing a command retrospectively by linking the identity and authentication information recorded on the blockchain with an audit trail. The blockchain-based transparency and audit module (240) can detect abnormal signs, such as abuse of authority or violation of authentication procedures, based on the results of the retrospective verification. The blockchain-based transparency and audit module (240) can additionally record the records of the retrospective verification on the blockchain to enable the prevention of repeated abuse and the tracking of responsibility.
[0124] In one embodiment, the blockchain-based transparency and audit module (240) can record the command propagation path and the arrival time for each relay node on the blockchain network. The blockchain-based transparency and audit module (240) can identify and record each communication path (satellite, mobile communication, internet, emergency frequency, etc.) through which the kill-switch command is propagated. The blockchain-based transparency and audit module (240) can record the unique identifier of each relay node and the command arrival time in the form of a timestamp. The blockchain-based transparency and audit module (240) can perform post-analysis of command delivery delays, path failures, and retransmission status through the recording of the propagation path and arrival time. The blockchain-based transparency and audit module (240) can utilize the propagation path recording as evidence for cause analysis in the event of a network failure or malicious attack.
[0125] In one embodiment, the blockchain-based transparency and audit module (240) can transparently manage the propagation history of a kill-switch command, including the command propagation path and the time of arrival at the node. The blockchain-based transparency and audit module (240) can visualize the entire transmission path of the command by sorting the propagation history data in chronological order. Through the transparent management of the propagation history, the blockchain-based transparency and audit module (240) can quickly detect abnormal situations such as command omissions, delays, or tampering. By ensuring the transparency of the propagation history, the blockchain-based transparency and audit module (240) can provide objective evidence in the event of international or legal disputes.
[0126] In one embodiment, the blockchain-based transparency and audit module (240) can record the execution results and release records of the target robot on a blockchain network. The blockchain-based transparency and audit module (240) can record execution result data, such as the time of completion of the robot's stop, the stop status, whether the safety stop sequence was executed, and the time of release approval, in standardized fields. The blockchain-based transparency and audit module (240) can include detailed information in the execution result data, such as the robot's sensor status, whether energy was cut off, and whether communication was cut off. The blockchain-based transparency and audit module (240) can make it impossible to manipulate or delete the execution results and release records afterward by storing them on the blockchain. The blockchain-based transparency and audit module (240) can objectively verify the actual execution and effect of the kill-switch command through the execution result records.
[0127] In one embodiment, the blockchain-based transparency and audit module (240) can manage the execution history of a kill-switch command, including execution results (time of completion of stop, confirmation of stop status, etc.). The blockchain-based transparency and audit module (240) can track the entire process of command execution by accumulating and recording execution history data in chronological order. Through execution history management, the blockchain-based transparency and audit module (240) can quickly detect abnormal situations such as non-compliance with commands, partial execution, or malfunctions. Through transparent management of execution history, the blockchain-based transparency and audit module (240) can provide objective evidence in the event of a dispute.
[0128] In one embodiment, the blockchain-based transparency and audit module (240) can record the reason for activation of the kill-switch, the issuer's identity information, authentication records, command propagation path, arrival time for each relay node, and the execution results and release records of the target robot on the blockchain network along with timestamps and digital signatures. The blockchain-based transparency and audit module (240) can assign a creation time to each record item as a timestamp in milliseconds. The blockchain-based transparency and audit module (240) can guarantee the authenticity and integrity of the data by combining the digital signature of the issuer or the system with each record item. The blockchain-based transparency and audit module (240) can strengthen the prevention of tampering with records and the legal evidentiary capacity through the combination of timestamps and digital signatures. The blockchain-based transparency and audit module (240) can enable consistent data management and rapid retrieval by grouping all record items into blockchain transaction units.
[0129] In one embodiment, the blockchain-based transparency and audit module (240) can secure tamper-proofing and legal evidentiary value by combining timestamps and electronic signatures for all audit records. Through the combination of timestamps and electronic signatures, the blockchain-based transparency and audit module (240) can clarify the time of record creation and the author. Due to the tamper-proofing effect, the blockchain-based transparency and audit module (240) can be used as reliable evidence in the event of a dispute. To secure legal evidentiary value, the blockchain-based transparency and audit module (240) can apply an international standard electronic signature algorithm.
[0130] In at least one embodiment of the present disclosure, the blockchain-based transparency and audit module (240) can transparently manage the entire process of a kill-switch command by uploading the record items to a blockchain network. The blockchain-based transparency and audit module (240) can create a transaction for each record item and then upload it to a distributed node of the blockchain network. The blockchain-based transparency and audit module (240) can verify the integrity of the data through a network consensus algorithm (e.g., PoW, PoS, etc.) at the time of upload. The blockchain-based transparency and audit module (240) can ensure that the uploaded records are synchronized with all participating nodes and managed transparently without a single point of failure. The blockchain-based transparency and audit module (240) can enable real-time auditing and post-verification through the automation of the upload procedure.
[0131] In at least one embodiment of the present disclosure, the blockchain-based transparency and audit module (240) may provide an audit system including an objection and compensation procedure for the unjust activation of a kill switch through uploading to a blockchain network. The blockchain-based transparency and audit module (240) may automate each stage of the audit system, such as receiving objections, reviewing, determining, and calculating compensation, based on the uploaded records. The blockchain-based transparency and audit module (240) may additionally record the details of the compensation procedure execution on the blockchain to provide results transparently to both the victim and the issuer. Through the automation of the audit system, the blockchain-based transparency and audit module (240) may secure the effect of rapid relief for unjust activation and prevention of recurrence.
[0132] In one embodiment, the blockchain-based transparency and audit module (240) can verify the legitimacy of a kill-switch command retrospectively through a blockchain network upload procedure. The blockchain-based transparency and audit module (240) can verify the legality of the issuance of the command and the appropriateness of its execution retrospectively by linking all uploaded records with an audit trail. The blockchain-based transparency and audit module (240) can prevent repeated abuse and system malfunctions by additionally recording the results of the retrospective verification on the blockchain. The blockchain-based transparency and audit module (240) can provide the legitimacy verification history in an international standard format so that it can be used as objective evidence in the event of inter-state disputes or legal disputes.
[0134] FIG. 3 is a drawing for explaining the detailed operation of each module within a robot according to one embodiment.
[0135] Referring to FIG. 3, in one embodiment, the robot (120) can determine its current state in real time based on its own sensor data. The robot (120), which is a Physical AI robot with a physical entity that interacts directly with the real-world environment, can periodically acquire various physical and environmental data, such as position, speed, acceleration, state of moving parts, battery level, temperature, and humidity, through a built-in sensor module. The robot (120) can preprocess the acquired sensor data in real time in a signal processing circuit. The robot (120) can classify the current operating state (movement, work, charging, etc.) by analyzing the preprocessed sensor data in a built-in processor. The robot (120) can minimize the possibility of false detection by filtering out abnormal values or noise in the sensor data. The robot (120) can store the results of the real-time state determination in internal memory or transmit them to an external system via wireless communication. The robot (120) can immediately detect sudden changes in state or danger signals through the real-time state determination function. The robot (120) can be used as a trigger for subsequent safety control operations based on real-time status determination results. In one embodiment, the robot (120) is a terminal subject to identity authentication of the present invention and may refer to a Human AI Robot, Embodied AI, Physical AI, or Humanoid Robot that has a physical entity and interacts directly with the real-world environment, rather than a simple software-based bot or virtual agent.Specifically, the AI robot in the present invention refers to a Physical AI robot capable of performing actions such as movement, manipulation, and interaction within a physical environment while having a body structure similar to that of a human, and is characterized by having a physical drive unit (actuator, joint, walking mechanism, etc.), a multi-sensor module (visual, tactile, auditory, distance sensor, etc.), and an Embodied AI function that enables physical interaction with the environment. Such a Physical AI robot can be implemented in various forms, such as autonomous robots, industrial humanoid robots, medical robots, service robots, military robots, flying drones, and underwater robots, and the kill-switch standard protocol of the present invention can be integrally applied to all such types of Physical AI robots.
[0136] In at least one embodiment of the present disclosure, the robot (120) can classify various operational states, such as movement, work, and charging, through real-time state monitoring. The robot (120) can classify multiple operational states, such as movement state, work state, charging state, standby state, and error state, according to predefined criteria. The robot (120) can determine the state by combining sensor data thresholds, movement patterns, location information, etc., for each operational state. If the robot (120) is classified as a movement state, it can track the movement path based on position change and speed data. If the robot (120) is classified as a work state, it can analyze the motor drive signal of the work area and the location information of the work tool. If the robot (120) is classified as a charging state, it can check the battery charging current and voltage and whether it is connected to a charging station. The robot (120) can utilize the state classification result as a conditional branching point for a safety stop sequence.
[0137] In at least one embodiment of the present disclosure, the robot (120) can utilize real-time status information for executing and reporting a safety stop sequence. The robot (120) can input real-time status information as an execution condition for the safety stop sequence. The robot (120) can determine a branching path for a customized safety stop sequence based on the status information. The robot (120) can automatically report the status information and the results of the safety stop sequence execution to an external registration agency or a central control system. The robot (120) can ensure transparency by linking the reported status information and execution results to a blockchain-based audit record. The robot (120) can support an external system in remotely monitoring the safety status of the robot through real-time reporting of status information.
[0138] In at least one embodiment of the present disclosure, the robot (120) can safely cut off control by executing a customized 'safety stop sequence' tailored to the current state to prevent secondary damage, such as falling or collision, caused by a sudden power cutoff. The robot (120) can select a sequence corresponding to the state from a predefined safety stop sequence table using real-time current state information as an input value. The robot (120) can control the operation of each drive unit step by step according to the selected safety stop sequence. The robot (120) can sequentially stop or cut off motors, actuators, power supply units, communication modules, etc., through control signals. The robot (120) can verify whether the stop operation has been successfully completed by checking sensor feedback at each step. The robot (120) can evaluate the possibility of secondary damage occurring in real time during the execution of the safety stop sequence. After the safety stop sequence is completed, the robot (120) can automatically report the fact of the control cutoff and the time of completion to an external system. The robot (120) can minimize secondary damage such as falling, collision, mechanical damage, and casualties by applying a customized safety stop sequence.
[0139] In at least one embodiment of the present disclosure, the robot (120) may apply a state-specific safety stop sequence, such as moving to a safe zone and stopping when moving, descending to a safe height and stopping when working at a height, withdrawing a surgical tool to a safe location and stopping when performing a medical procedure, and moving to a safe roadside and stopping when transporting. When the robot (120) is classified as moving, it may search for the nearest safe zone using built-in map data and a position sensor. The robot (120) may calculate a path to the safe zone and move while gradually reducing its speed. When the robot (120) is classified as working at a height, it may check the height information of the work area and lower the work platform at a safe descent speed. When the robot (120) is classified as performing a medical procedure, it may move the surgical tool to a location outside the patient's body and secure the tool to a safe location. When the robot (120) is classified as transporting, it may move to the edge of the road or a designated waiting area and stop. After executing the state-specific sequence, the robot (120) may verify with a sensor whether the stop is complete.
[0140] In at least one embodiment of the present disclosure, the robot (120) may immediately perform minimum safety measures (energy cutoff, drive unit locking) in emergency situations where the safety stop sequence cannot be completed. If the robot (120) determines, based on sensor data or internal diagnostic results, that the normal completion of the safety stop sequence is impossible, it may immediately cut off the energy supply. The robot (120) may cut off the power to the drive unit and, at the same time, operate a mechanical locking device to fix all joints or moving parts. After executing the minimum safety measures, the robot (120) may immediately report the fact of the emergency stop and the cause code to an external system. By executing the minimum safety measures, the robot (120) can prevent the spread of secondary damage in emergency situations, such as unexpected malfunctions, communication loss, or hardware failure.
[0141] In one embodiment, the robot (120) can define different safety stop sequences for each type of robot (mobile robot, stationary industrial robot, flying drone, underwater robot, medical robot, service robot). The robot (120) can embed a safety stop sequence table optimized for each type based on pre-registered Physical AI robot type information. The robot (120) can apply separate sequence logic to various forms of Physical AI robots, such as humanoid robots, embodied AI-based mobile robots, stationary industrial robots, flying drones, underwater robots, medical robots, and service robots. The robot (120) for each robot type
[0142] In one embodiment, the robot (120) can stop by gradually reducing its speed and guiding it to a safe zone in the case of a mobile robot. If the robot (120), which is a Physical AI robot, is classified as a mobile robot, a deceleration curve can be applied through a speed control algorithm, and in particular, if it is a humanoid robot, a walking deceleration algorithm can be applied to gradually reduce the walking speed while maintaining bipedal balance. During the deceleration process, the robot (120) can monitor the surrounding environment in real time using an obstacle detection sensor. Upon reaching the safe zone, the robot (120) can stop the drive motor and activate the parking brake or mechanical lock. After the stop is complete, the robot (120) can report location information and the stopped status to an external system.
[0143] In at least one embodiment of the present disclosure, the robot (120) can descend to and land at an emergency landing point in the case of a flying drone. The robot (120), which is an Embodied AI-based Physical AI robot, can locate an emergency landing point by utilizing an embedded GPS and altitude sensor when classified as a flying drone. The robot (120) can calculate an optimal path to the emergency landing point and synchronize the descent speed and attitude control. The robot (120) can correct wind, obstacle, and terrain information in real time during the landing process. After landing is complete, the robot (120) can completely stop the propeller and propulsion motor.
[0144] In at least one embodiment of the present disclosure, the robot (120) can fix its position after surfacing in the case of an underwater robot. If the robot (120), which is a Physical AI robot, is classified as an underwater robot, it can calculate an ascent path to the surface using a depth sensor and a position sensor by utilizing an Embodied AI-based environment perception function.
[0145] In one embodiment, the robot (120), in the case of a medical robot, can switch to standby mode after safely withdrawing the surgical tool from the patient's body. If the robot (120), which is a Physical AI robot, is classified as a medical humanoid robot or a medical robot, it can measure the location of the surgical tool and the distance from the patient's body in real time. The robot (120) can generate a control signal to move the surgical tool to a safe location outside the patient's body. After the withdrawal of the tool is complete, the robot (120) can switch all actuators to standby mode and block further operation. The robot (120) can report the fact of switching to standby mode and the tool location information to an external system.
[0146] In at least one embodiment of the present disclosure, the robot (120) may immediately switch to a Level 4 (energy cutoff) measure if the maximum allowable time of each type of safety stop sequence is exceeded. When executing the safety stop sequence, the robot (120) may measure the elapsed time for each step. If the stop sequence is not completed within a predefined maximum allowable time, the robot (120) may immediately switch to a Level 4 measure to cut off the energy supply. After executing the Level 4 measure, the robot (120) may forcibly stop all operations of the drive unit. The robot (120) may report the fact of switching to the Level 4 measure and the cause code to an external system.
[0147] A separate dedicated security chip (Secure Element) may be provided. The robot (120), which is a physical AI robot, may embed a dedicated security chip (Secure Element) in a circuit area separate from the main operating system during the hardware design phase. The dedicated security chip may have a power supply and communication path independent of the robot's main processor. The dedicated security chip can operate independently even in abnormal situations such as hacking of the operating system, software errors, or system crashes. The dedicated security chip can independently handle the reception, verification, and execution of kill-switch commands. The independence of the dedicated security chip can significantly improve the reliability and security of command execution.
[0148] In at least one embodiment of the present disclosure, a dedicated security chip (Secure Element) mounted on robot hardware can receive and execute kill-switch commands by operating independently of the robot's main operating system. The dedicated security chip can directly receive kill-switch commands from a network through a separate communication receiver. The dedicated security chip can independently perform quadruple verification of the received command, including digital signature, authority, validity period, and geographical range. The dedicated security chip can control the operation of the robot through a drive unit control interface only for commands that have passed verification. The dedicated security chip can independently record the results of command execution and report them to an external system.
[0149] In at least one embodiment of the present disclosure, a dedicated security chip can prevent unauthorized modification of firmware through hardware-based secure boot. When power is applied, the dedicated security chip can verify the digital signature of the firmware through a built-in secure bootloader. The dedicated security chip can block the execution of firmware that fails signature verification. If the dedicated security chip detects an attempt at unauthorized modification, it can immediately automatically trigger a kill-switch command or switch the system to safe mode. Hardware-based secure boot can fundamentally block security threats such as firmware tampering and malicious code insertion.
[0150] In at least one embodiment of the present disclosure, the dedicated security chip may include an independent communication receiver, a command verification unit, and a drive unit control interface. The communication receiver of the dedicated security chip may be directly connected to an external network to receive commands. The command verification unit may incorporate quadruple verification logic, such as the digital signature, authority, validity period, and geographical range of the received command. The drive unit control interface may directly control key hardware, such as a motor, actuator, and power supply unit, according to the verified command. Each module may be hardware-separated to prevent mutual interference.
[0151] In at least one embodiment of the present disclosure, the dedicated security chip may be designed to be physically embedded during robot manufacturing so that it cannot be removed afterwards. The dedicated security chip may be physically embedded by methods such as soldering, epoxy sealing, or hardware encryption on the main board of the robot body or a separate security module. The dedicated security chip may have a hardware defense structure to prevent it from being arbitrarily separated, replaced, or dismantled from the outside. The physical embedding of the dedicated security chip can prevent hardware attacks such as removal, forgery, or duplication of the security chip.
[0152] In one embodiment, firmware updates for the security chip may be permitted only when there is a joint digital signature from the manufacturer and the registration authority. The dedicated security chip can verify both pairs of digital signatures from the manufacturer and the registration authority upon a firmware update request. The dedicated security chip can proceed with the firmware update only when signature verification is complete. The dedicated security chip can record update history and signature information in internal memory. The dedicated security chip can automatically block unauthorized update attempts.
[0153] In at least one embodiment of the present disclosure, a kill switch can be automatically activated when an unauthorized update attempt is detected. A dedicated security chip can detect unauthorized firmware modification or unauthorized update attempts in real time. Upon detection of an unauthorized attempt, the dedicated security chip can cut off the energy supply to the robot or switch the drive unit to a locked state. The dedicated security chip can automatically report the fact of the unauthorized update attempt and related logs to an external system. The dedicated security chip can prevent the spread of security threats in advance through the automatic activation of the kill switch.
[0154] In at least one embodiment of the present disclosure, the dedicated security chip can receive and execute a kill-switch command even when the robot's main operating system is completely disabled. The dedicated security chip can operate independently of the operating system through its own power supply and independent communication path. The dedicated security chip can receive and execute commands even in abnormal situations, such as operating system boot failure, system crash, or malware infection. The dedicated security chip can output hardware control signals directly without exchanging data with the operating system. The independent operation of the dedicated security chip can maximize the security and reliability of the entire robot.
[0155] In one embodiment, the robot (120) may include a safety restart protocol module (270) that performs multi-stage safety verification in the order of hardware integrity self-diagnosis, software normal state verification, sensor calibration and environment recognition function verification, test run within a limited range, and stepwise restoration of operational authority upon release of the kill switch. Upon receiving a kill switch release command, the robot (120) may perform a hardware integrity self-diagnosis as the first stage. In the hardware self-diagnosis, the robot (120) may check whether key components (sensors, motors, power supply units, etc.) are operating normally. In the second stage, the robot (120) may check the normal state of the software and AI models. In the software diagnosis, the robot (120) may check firmware integrity, normal execution of key processes, and whether there is a malicious code infection. In the third stage, the robot (120) may perform sensor calibration and environment recognition function verification. During the sensor calibration process, the robot (120) may correct the error between the reference value and the measured value. The robot (120) can evaluate the safety of surrounding obstacles, location, and work environment in the environment recognition verification. The robot (120) can verify the safety of actual operation by conducting a test run within a limited range in four stages. The robot (120) can determine whether there are any abnormalities by analyzing the results of the test run. The robot (120) can be configured to enable full operation by gradually restoring operational authority in five stages. Verification at each stage can be automated based on a predefined checklist and threshold values.
[0156] In at least one embodiment of the present disclosure, the robot (120) can automatically report the verification results of each step in a multi-step safety restart procedure to a registration authority. Upon completion of each verification step, the robot (120) can transmit result data (diagnostic results, presence or absence of abnormalities, calibration values, etc.) in real time to a registration authority server. The robot (120) can receive approval or a request for additional inspection from the registration authority regarding the reported verification results. The robot (120) can ensure transparency by linking the verification result report history to a blockchain-based audit record.
[0157] In one embodiment, the robot (120) may allow full operation to resume only when it has passed all verification steps. The robot (120) may proceed to the next step only when the verification result of each step is determined to be normal. The robot (120) may resume operation of all functions, such as the drive unit, communication unit, and work unit, only after all verification steps have been completed. If the verification is not passed, the robot (120) may switch to a restricted operation or additional inspection mode.
[0158] In at least one embodiment of the present disclosure, the robot (120) can prevent additional risks resulting from improper restart through multi-stage safe restart. The robot (120) can prevent potential risk factors, such as hardware defects, software errors, and environmental risks, in advance through a multi-stage verification procedure. The robot (120) can prevent additional risks, such as secondary accidents, malfunctions, and casualties, by restricting restart when verification is not passed. The robot (120) can improve the reliability and safety of the entire system by applying a safe restart protocol.
[0159] In at least one embodiment of the present disclosure, the robot (120) can periodically acquire various physical and environmental data, such as position, velocity, acceleration, state of the moving part, remaining battery level, temperature, and humidity, through a built-in sensor module. The robot (120) can acquire real-time position coordinates through a position sensor. The robot (120) can measure movement velocity and acceleration values through a velocity sensor and an acceleration sensor. The robot (120) can use multiple sensors, such as a joint angle sensor, a motor current sensor, and an encoder, to detect the state of the moving part. The robot (120) can measure the charge status of the battery through a voltage sensor and a current sensor to check the remaining battery level. The robot (120) can acquire temperature and humidity values of the surrounding environment through a temperature sensor and a humidity sensor. The robot (120) can automatically adjust the collection cycle of sensor data to a preset cycle (e.g., 100ms, 1 second, etc.). The robot (120) can automatically shorten the sensor data collection cycle when an operating state or danger signal is detected, thereby enabling a rapid response to abnormal situations. The robot (120) can synchronize multiple sensor data to accurately determine the state at the same point in time. When an abnormal signal or defect occurs in the sensor module, the robot (120) can maintain the continuity of data acquisition through a backup sensor or an alternative algorithm. The robot (120) can temporarily store the acquired sensor data in internal memory or transmit it to a signal processing circuit in real time. The robot (120) can record the process of acquiring and storing sensor data in a log file, which can be used for post-analysis and audit records.
[0160] In one embodiment, the robot (120) can preprocess the acquired sensor data in real time in a signal processing circuit. The robot (120) can convert the analog signal of the sensor data into a digital signal in the signal processing circuit. The robot (120) can apply various signal processing algorithms, such as a low-pass filter, a Kalman filter, and a moving average filter, to remove noise generated during the sampling process of the sensor data. The robot (120) can improve the accuracy of data analysis by performing time synchronization and normalization of the sensor data. The robot (120) can automatically detect missing values or abnormal values during the real-time preprocessing process. The robot (120) can evaluate the quality of the preprocessed data and exclude data with low reliability from subsequent analysis. The robot (120) can transmit the preprocessed sensor data to an internal processor or an external analysis module. In one embodiment, the robot (120) can minimize the possibility of false detection by filtering out abnormal values or noise in the sensor data. The robot (120) can determine sensor data exceeding a predefined threshold as an abnormal value. If an outlier is detected, the robot (120) can ignore or re-measure the corresponding data. The robot (120) can automatically identify data segments containing noise and adjust the reliability weight. The robot (120) can use the noise filtering results as input values for state classification and safety stop sequences. By minimizing the possibility of false detection, the robot (120) can prevent unnecessary safety stops or malfunctions. The robot (120) can continuously maintain the reliability of state recognition by repeatedly performing data preprocessing and noise filtering processes in real time.
[0161] In at least one embodiment of the present disclosure, the robot (120) can classify the current operating state (movement, work, charging, etc.) by analyzing preprocessed sensor data in an embedded processor. The robot (120) can classify the state into one of a plurality of operating states, such as a movement state, a work state, or a charging state, by analyzing the pattern of the sensor data in the embedded processor. For state classification, the robot (120) can analyze a combination of multiple sensor data, such as position change, speed, acceleration, motor drive signal, and battery charging state. The robot (120) can implement a state classification algorithm based on predefined rules or machine learning. The robot (120) can update the state classification results in real time to quickly reflect changes in the operating state. In at least one embodiment of the present disclosure, the robot (120) can classify a plurality of operating states, such as a movement state, a work state, a charging state, a standby state, and an error state, according to predefined criteria. The robot (120) can determine the movement state based on a position change and whether a speed threshold is exceeded. The robot (120) can determine the working state by the activation of the motor drive signal and the change in the position of the work tool. The robot (120) can determine the charging state by the battery charging current, voltage, and charging station connection signal. The robot (120) can determine the standby state by the stoppage of all drive parts and energy consumption patterns. The robot (120) can determine the error state according to predefined error conditions such as sensor defects, communication failures, and abnormal operation patterns. In one embodiment, the robot (120) can determine the state by combining the threshold value of sensor data, operation patterns, and position information for each operation state. The robot (120) can detect a state transition when the sensor data deviates from a specific range through a threshold-based determination logic. The robot (120) can identify repetitive or abnormal movements through operation pattern analysis.The robot (120) can track a movement path and a stopping position by combining location information and time information. In at least one embodiment of the present disclosure, when the robot (120) is classified as being in a moving state, it can track a movement path based on location change and speed data. The robot (120) can utilize the results of tracking the movement path for predicting the path of a safety stop sequence and avoiding dangerous zones. The robot (120) can predict the risk of collision in advance by further analyzing obstacle detection sensor data while moving. In one embodiment, when the robot (120) is classified as being in a working state, it can analyze the motor drive signal of the work area and the location information of the work tool. The robot (120) can detect overload or abnormal operation by analyzing additional sensor data such as motor current, torque, and temperature of the work area. The robot (120) can monitor changes in the location of the work tool and changes in the state of the work target in real time. In at least one embodiment of the present disclosure, when the robot (120) is classified as being in a charging state, it can check the battery charging current and voltage and whether a charging station is connected. The robot (120) can additionally check the charging speed, the predicted time to complete charging, and whether the charging station is operating normally during the charging state. If an abnormal signal occurs during charging, the robot (120) can output a charging stop or warning signal. In one embodiment, the robot (120) can use the state classification result as a conditional branching point for a safety stop sequence. Depending on the state classification result, the robot (120) can select a customized safety stop sequence, such as stopping while moving, stopping while working, or stopping while charging. The robot (120) can report the state classification result to an external system to be used as an input value for remote control and monitoring.
[0162] In one embodiment, the robot (120) can store the real-time status identification results in internal memory or transmit them to an external system via wireless communication. The robot (120) can record real-time status information in sequential data blocks of internal memory. The robot (120) can automatically adjust the storage cycle and data retention period of the status information according to preset values. The robot (120) can transmit the status information to an external system, such as a central control system, a registration agency, or a manufacturer's server, via a wireless communication module. When transmitting status information, the robot (120) can perform data encryption and integrity verification procedures. In the event of a communication failure, the robot (120) can temporarily store the status information and transmit it in batches upon communication recovery. In one embodiment, the robot (120) can input real-time status information as an execution condition for a safety stop sequence. The robot (120) can automatically trigger the execution of a safety stop sequence when a change in the status information satisfies a predefined threshold condition. The robot (120) can monitor real-time changes in status information and immediately execute a safety stop sequence upon detection of a danger signal. In at least one embodiment of the present disclosure, the robot (120) can determine a branch path of a customized safety stop sequence based on the status information. The robot (120) can refer to different safety stop sequence tables depending on the type of status information (movement, operation, charging, error, etc.). The robot (120) can store the branch path determination result in internal logic and use it as a basis for subsequent actions. In one embodiment, the robot (120) can automatically report the status information and the safety stop sequence execution result to an external registration agency or a central control system. When automatically reporting, the robot (120) can adjust the reporting cycle, reporting target, and reporting format according to preset values. The robot (120) may include a timestamp, electronic signature, status classification code, execution result code, etc. in the reported data.In at least one embodiment of the present disclosure, the robot (120) can ensure transparency by linking reported status information and execution results to a blockchain-based audit record. The robot (120) can upload status information and execution results to a blockchain network in the form of transactions. Through the blockchain record, the robot (120) can transparently verify the execution history of kill-switch commands and changes in status. In one embodiment, the robot (120) can support an external system in remotely monitoring the safety status of the robot through real-time reporting of status information. The robot (120) can respond in real-time to a remote monitoring request from an external system. Through the real-time reporting function, the robot (120) can induce immediate intervention by an external system when an abnormal state occurs.
[0163] In one embodiment, the robot (120) can determine its current state in real time based on its own sensor data upon receiving a kill-switch command. The robot (120) can collect multiple sensor data in real time, such as a built-in position sensor, speed sensor, acceleration sensor, joint angle sensor, and battery sensor. The robot (120) can preprocess the collected sensor data in a signal processing circuit to remove noise and evaluate the reliability of the data. The robot (120) can classify the current operating state, such as moving, working, charging, or standing by analyzing the preprocessed sensor data in a built-in processor. Based on the state classification results, the robot (120) can identify in advance the expected risk factors for each state (e.g., falling while moving, collision while working, overcurrent while charging, etc.). By analyzing the risk factors for each state, the robot (120) can evaluate the type and impact of secondary damage (falling, collision, mechanical damage, etc.) that may occur in the event of a sudden power cut. The robot (120) can use the results of the risk factor analysis as input values for the design of a safety stop sequence. This real-time status identification and risk factor analysis enables the rapid application of a customized safety stop sequence when executing a kill-switch command.
[0164] For example, when moving, the speed can be gradually reduced, and the robot can move to a safe zone and then stop. In particular, in the case of a humanoid robot, when walking, the robot can maintain balance and transition to a safe posture (such as sitting or kneeling) before stopping. When working, the robot can move the work tool to a safe position and stop the moving part. When charging, the robot can safely cut off the charging circuit and stop the energy supply. The robot (120) can monitor the execution results of each step in real time using sensor data to verify whether the stop sequence is proceeding normally. The robot (120) can indicate whether the stop sequence is completed using an internal flag and can safely cut off control upon completion. Execution of such a customized safety stop sequence can effectively prevent secondary damage caused by sudden cutoff.
[0165] In at least one embodiment of the present disclosure, the robot (120) can minimize secondary damage caused by sudden shutdown by applying a safe stop sequence according to state. The robot (120) can define different safe stop sequences for each operation state, such as movement, work, charging, and standby. The robot (120) can analyze in advance the types of secondary damage for each state, such as falling while moving, collision while working, and overcurrent while charging. The robot (120) can include step-by-step control commands (speed reduction, tool withdrawal, energy cutoff, etc.) to prevent secondary damage for each state in the safe stop sequence. When executing the safe stop sequence, the robot (120) can verify the success of each step in real time using sensor data. The robot (120) can record the results of the step-by-step execution of the stop sequence as an internal log and use it for post-audit and analysis. By applying a strategy to minimize secondary damage, the robot (120) can significantly reduce incidental risks such as human casualties, property damage, and mechanical damage.
[0166] In at least one embodiment of the present disclosure, the robot (120) may immediately perform minimum safety measures (energy cutoff, drive unit locking) in emergency situations where the completion of the safety stop sequence is impossible. The robot (120) may pre-set the maximum allowable time for each step of the safety stop sequence. The robot (120) may immediately switch to minimum safety measures if the stop sequence is not completed normally or if an emergency situation (sensor malfunction, communication loss, mechanical failure, etc.) is detected. As a minimum safety measure, the robot (120) may stop the operation of all drive units by cutting off the energy supply circuit. The robot (120) may activate the drive unit locking device to prevent physical movement or malfunction. The robot (120) may record the results of the execution of the minimum safety measures in internal flags and logs. The immediate execution of these minimum safety measures provides the effect of quickly preventing secondary damage even when the stop sequence is not completed.
[0167] In one embodiment, when the robot (120) is moving, it can gradually reduce its speed and guide itself to a safe zone to stop. When the robot (120) is classified as being in a moving state, it can gradually reduce its current speed through a built-in speed control algorithm. The robot (120) can locate a predefined safe zone (e.g., a stopping zone, an emergency shelter, etc.) by utilizing built-in position sensors and map data. The robot (120) can calculate the optimal path from the current location to the safe zone through a path planning algorithm. The robot (120) can continue moving toward the safe zone while simultaneously decelerating. After reaching the safe zone, the robot (120) can output a complete stop signal and set the output of the drive motor to zero. After stopping is complete, the robot (120) can cut off control and automatically report status information to a central system. This step-by-step execution provides the effect of preventing secondary damage, such as falling or collision, caused by sudden stopping while moving.
[0168] In at least one embodiment of the present disclosure, the robot (120) may descend to a safe height and stop when working at a height. When the robot (120) is classified as working at a height, it may measure its current height in real time through a built-in altitude sensor and position control algorithm. The robot (120) may execute a descent command to descend to a predefined safe height (e.g., 1m or less above the ground). The robot (120) may apply a speed control algorithm during the descent process to prevent a sudden fall. After reaching the safe height, the robot (120) may cut off the drive unit output to come to a complete stop. After the stop is complete, the robot (120) may automatically report status information to a central system. This stepwise descent and stop execution provides the effect of preventing fall accidents caused by sudden stopping during work at a height.
[0169] In at least one embodiment of the present disclosure, the robot (120) may switch to a standby mode after safely withdrawing the surgical tool from the patient's body when a medical procedure is in progress. When the robot (120) is classified as being in a medical procedure state, it may check the current location of the surgical tool in real time through a built-in tool location sensor and control algorithm. The robot (120) may execute a withdrawal command to move the surgical tool to a predefined safe location outside the patient's body. The robot (120) may precisely control the speed and path during the tool withdrawal process to prevent contact or damage to the patient. After the tool reaches the safe location, the robot (120) may switch to a standby mode and cut off all drive unit outputs. The robot (120) may automatically report the results of the stop and standby mode transition to a central system. This stepwise tool withdrawal and standby mode transition provides the effect of preventing patient damage caused by sudden interruption during a medical procedure.
[0170] In one embodiment, if the robot (120) is a flying drone, it can descend to an emergency landing point and land. When the robot (120) is classified as being in a flight state, it can measure its current location and altitude in real time through a built-in GPS, altitude sensor, and obstacle detection sensor. The robot (120) can locate a predefined emergency landing point (e.g., an emergency landing pad, an open area, etc.). The robot (120) can calculate the optimal descent path from its current location to the emergency landing point through a path planning algorithm. The robot (120) can land safely by applying an obstacle avoidance algorithm during the descent process. After landing is complete, the robot (120) can cut off the output of all propulsion motors and automatically report status information to a central system. This stepwise descent and landing execution provides the effect of preventing crash accidents caused by sudden cutoff during flight.
[0171] In one embodiment, if the robot (120) is an underwater robot, it can fix its position after surfacing to the water surface. If the robot (120) is classified as being underwater, it can measure the current water depth in real time through a built-in depth sensor and position control algorithm. The robot (120) can gradually ascend to the water surface by executing a predefined surfacing command. The robot (120) can precisely control propulsion and direction during the surfacing process to prevent sudden surfacing or collision. After reaching the water surface, the robot (120) can execute a position fixing command to minimize movement caused by external currents or waves. The robot (120) can automatically report the results of surfacing and position fixing to a central system. This stepwise surfacing and position fixing execution provides the effect of preventing secondary damage, such as flooding, loss, or mechanical damage caused by the sudden blockage of the underwater robot.
[0172] In at least one embodiment of the present disclosure, the robot (120) may immediately switch to a Level 4 (energy cutoff) measure if the maximum allowable time of each type of safety stop sequence is exceeded. The robot (120) may pre-set the maximum allowable time for each step of the safety stop sequence. If the stop sequence is not completed within the maximum allowable time, the robot (120) may automatically execute a Level 4 (energy cutoff) measure command. The robot (120) may cut off the energy supply circuit to immediately stop the operation of all drive units. The robot (120) may record the energy cutoff result in an internal flag and log and automatically report it to a central system. This immediate switching to a Level 4 measure provides the effect of quickly preventing secondary damage even when the stop sequence is not completed.
[0173] In at least one embodiment of the present disclosure, the robot (120) can immediately perform minimum safety measures (energy cutoff, drive unit locking) in emergency situations where the safety stop sequence cannot be completed. When an emergency situation is detected, such as a sensor malfunction, communication loss, or mechanical failure, the robot (120) can skip the remaining steps of the safety stop sequence and immediately switch to minimum safety measures. The robot (120) can stop the operation of all drive units by cutting off the energy supply circuit. The robot (120) can prevent physical movement or malfunction by activating the drive unit locking device. The robot (120) can record the results of the execution of minimum safety measures in internal flags and logs and automatically report them to a central system. The immediate execution of these minimum safety measures provides the effect of quickly preventing secondary damage in emergency situations.
[0174] In at least one embodiment of the present disclosure, the robot (120) may be equipped with a dedicated security chip (Secure Element) that is physically separated from the main operating system (OS) so that commands can be reliably executed even in the event of hacking or an operating system error. The robot (120) may place the dedicated security chip (Secure Element) in a separate circuit area on the main board. The dedicated security chip may operate through a separate power supply line from the main processor on which the main operating system (OS) runs. The dedicated security chip separates the data bus and memory access path from the main operating system at the hardware level so that command reception and execution may not be affected even if OS hacking or software errors occur. The dedicated security chip may be designed to be directly soldered to the main board or embedded via a chip-on-board method during robot manufacturing so that it cannot be physically removed or bypassed afterwards. The dedicated security chip is responsible only for core security functions such as receiving, verifying, and controlling execution of kill-switch commands, and communication with the main operating system may be allowed only through strictly limited interfaces (e.g., unidirectional signal lines, hardware gateways). This physical isolation structure ensures that kill-switch commands can be received and executed independently, even in situations where the main operating system is hacked or operates abnormally. Through the application of a dedicated security chip, the reliability and security of standardized forced blocking protocols can be significantly enhanced, even in multi-robot environments with differing manufacturers, operating systems, and communication methods.
[0175] In at least one embodiment of the present disclosure, a dedicated security chip (Secure Element) mounted on robot hardware can prevent unauthorized modification of firmware through hardware-based secure boot. Whenever power is applied, the dedicated security chip allows the embedded secure bootloader to verify the digital signature or hash value of the firmware. The secure bootloader can trust only firmware authenticated by digital signatures from the manufacturer and registration authority. If unauthorized modification, tampering, or an attempt at an unauthorized update of the firmware is detected, the dedicated security chip can stop booting or generate a signal to automatically trigger a kill switch. Hardware-based secure boot can prevent not only software-level attacks (e.g., rootkits, viruses) but also attempts at hardware intrusion, such as firmware rollback attacks and bootloader tampering. The secure boot procedure is performed repeatedly at every boot time to continuously guarantee the integrity and reliability of the firmware. By applying such hardware-based secure boot, the security of the kill switch receiving and executing firmware is maximized, and the safety of the entire system can be improved.
[0176] In at least one embodiment of the present disclosure, the kill-switch receiving and execution firmware module (280) may perform quadruple verification of the digital signature of the command message, the issuer authority, the command validity period, and the geographical range to verify whether the kill-switch command is a legitimate command. The kill-switch receiving and execution firmware module (280) may first verify the digital signature of the received command message. The firmware module (280) may refer to an embedded trusted public key store to verify whether the signature of the command issuer matches the identity of the manufacturer or registration authority. The firmware module (280) may verify whether the authority corresponds to the blocking level by checking the issuer authority information included in the command message. The firmware module (280) may determine whether the current time falls within the validity period of the command by checking the validity period field of the command message. The firmware module (280) may verify whether the current location of the robot (120) falls within the coverage range of the command by checking the geographical range information of the command message. If any of the four verifications fails, the firmware module (280) may refuse to execute the kill-switch command and record the reason for refusal in an internal log. The firmware module (280) may execute the kill-switch command only when all four verifications have been passed. This four verification procedure can effectively block various security threats such as unauthorized command execution, abuse of authority, falsification of commands, and misuse of commands outside the local area.
[0177] In at least one embodiment of the present disclosure, the kill-switch receiving and executing firmware module (280) can prevent unauthorized modification of the firmware through hardware-based secure boot. The firmware module (280) can verify the digital signature or hash value of the firmware image through a built-in secure bootloader at each boot time. When unauthorized modification of the firmware is detected, the firmware module (280) can immediately stop the kill-switch command receiving and executing function. The firmware module (280) can record the unauthorized modification attempt in an internal log and automatically send a warning signal to the manufacturer and the registration authority. If unauthorized modification is repeatedly detected, the firmware module (280) can forcibly block the drive unit of the robot (120) or switch the security chip itself to a lock mode. This firmware unauthorized modification detection and blocking function can protect the system from various security threats, such as firmware tampering, unauthorized updates, and malicious code insertion.
[0178] In one embodiment, the kill-switch receiving and execution firmware module (280) may include an independent communication receiver, a command verification unit, and a drive unit control interface within a dedicated security chip so that it can receive and execute kill-switch commands even when the robot's main operating system is completely disabled. The kill-switch receiving and execution firmware module (280) may have a separate communication receiver embedded within the dedicated security chip to receive kill-switch commands directly from a network. The firmware module (280) can independently perform quadruple verification, such as digital signature, authority, validity period, and geographical range, of the received command through the command verification unit. The firmware module (280) can directly execute forced shutdown commands, such as power cutoff, motor locking, and communication cutoff, regardless of the main operating system (OS), through the drive unit control interface. The firmware module (280) can ensure the reception and execution of kill-switch commands by utilizing the independent hardware resources of the dedicated security chip even when the main operating system is hacked, erroneous, or completely disabled. The firmware module (280) records the command execution results as internal flags and logs, and can automatically report the execution results externally when necessary. This independent forced block execution structure enables the robot to ultimately secure safety control even in the event of software-based attacks or system failures.
[0179] In at least one embodiment of the present disclosure, the safety restart protocol module (270) according to the present invention can verify the safety of a Physical AI robot (humanoid robot, embodied AI robot, human AI robot, etc.) in multiple stages when the kill switch is released, and then restore the function step by step. The safety restart protocol module (270) is a kill switch
[0180] In at least one embodiment of the present disclosure, the safety restart protocol module (270) can automatically report the results of each verification step to a registration authority. Whenever verification of each step is completed, the safety restart protocol module (270) can convert the verification result data into a standardized message format. The converted verification results can be automatically transmitted to the registration authority's server via a network. The automatic reporting function can ensure transparency in the verification process and support real-time monitoring and post-audit by external agencies. The reported data may include verification steps, inspection items, result values, timestamps, robot identification numbers, etc. The automatic reporting procedure can be linked with a blockchain-based audit record system to prevent tampering with the verification history and enhance its legal evidentiary value.
[0181] In one embodiment, the safety restart protocol module (270) may allow full operation to resume only when all verification steps have been passed. The safety restart protocol module (270) may check whether each verification step has been passed in its internal logic. Only when all steps have been successfully completed may operational authority over the robot's full functions be restored. If an anomaly is detected in one or more steps, the robot may maintain only limited functions or postpone restart. These step-by-step passing conditions may provide the effect of maximizing safety during the restart process.
[0182] In at least one embodiment of the present disclosure, the safety restart protocol module (270) can perform a self-diagnosis of the hardware integrity of the robot in the first step. In the hardware integrity self-diagnosis step, the safety restart protocol module (270) can check the status of key hardware components through a built-in diagnostic algorithm. The self-diagnosis can verify whether the electrical signals, response speeds, temperatures, and current values of key components, such as power supplies, sensors, actuators, communication modules, and security chips, are within a normal range by measuring them. The diagnosis results are compared with predefined threshold values, and if abnormal signs are detected, an error code for the corresponding component may be generated. The hardware self-diagnosis can be automatically repeated by a diagnostic circuit or firmware inside the robot. The self-diagnosis results are stored separately by verification step and can be used as a criterion for determining whether to proceed to a subsequent step.
[0183] In one embodiment, hardware integrity self-diagnosis can check whether key hardware components, such as sensors, actuators, and communication modules, are operating normally. The normal operation of a sensor can be evaluated by measuring the range of sensor output values, response time, noise level, etc. The normal operation of an actuator can be checked in real time by monitoring the motor's rotational speed, torque, position feedback signal, etc. The normal operation of a communication module can be determined based on the network connection status, packet transmission and reception success rate, signal strength, etc. The inspection results of each hardware component are recorded individually, thereby minimizing the impact of a defect in a specific component on the entire system. The hardware integrity self-diagnosis step can provide the effect of preventing unsafe restarts by detecting malfunctions caused by physical defects or aging in advance.
[0184] In at least one embodiment of the present disclosure, the safety restart protocol module (270) can perform a normal state check of the software and AI model in two stages. In the software normal state check stage, the safety restart protocol module (270) can check the integrity and normal operation of the operating system (OS), control software, and AI algorithm. The software integrity check can verify whether tampering has occurred by comparing the hash value or digital signature of each software module with a pre-registered reference value. The normal operation can be evaluated by analyzing system resource information such as the software's boot log, error log, process status, memory usage, and CPU occupancy rate. The normal state of the AI model can be checked based on the integrity of model parameters, input-output tests, and the reliability of prediction results. The software check results can be linked with hardware diagnostic results to ensure a consistent normal state of the entire system.
[0185] In one embodiment, the software normal state verification can check the integrity and normal operation of the operating system, control software, and AI algorithm. The integrity of the operating system can be verified through the verification of hash values of the kernel, drivers, and system libraries. The normal operation of the control software can be evaluated by analyzing the execution logs of key control routines, exception handling results, interface response times, etc. The integrity of the AI algorithm can be checked through the verification of digital signatures of model files, comparison of parameter checksums, and verification of the consistency of prediction results for test inputs. The software normal state verification step can provide the effect of preventing the risk of malfunction caused by malware infection, software errors, unauthorized changes, etc., in advance.
[0186] In at least one embodiment of the present disclosure, the safety restart protocol module (270) may perform sensor calibration and environment recognition function verification in three stages. In the sensor calibration stage, the safety restart protocol module (270) may correct errors by comparing the reference value of each sensor with the actual measured value. Sensor calibration may be performed individually for various sensors, such as temperature, distance, acceleration, gyroscope, and camera. Environment recognition function verification may be performed through test scenarios to confirm whether the robot accurately recognizes obstacles, boundaries, people, and other objects in the surrounding environment. During the verification process, the reliability of sensor data, recognition accuracy, response speed, false positive rate, etc., may be used as evaluation indicators. The results of the sensor calibration and environment recognition verification may serve as key criteria for determining whether the robot can operate safely in a real environment.
[0187] In at least one embodiment of the present disclosure, the environment perception function verification can confirm whether the robot accurately perceives the surrounding environment and ensures the reliability of the sensor data. The accuracy of the environment perception can be verified by having the robot perform multiple scenarios, such as obstacle avoidance, path tracking, and human recognition, in a predefined test environment. The reliability of the sensor data can be evaluated by analyzing data consistency, noise levels, signal loss rates, etc., during repeated measurements in the same environment. The environment perception function verification step can provide the effect of preventing unexpected accidents caused by sensor malfunctions or recognition errors.
[0188] In one embodiment, the safety restart protocol module (270) can restore operational authority in stages. The safety restart protocol module (270) can gradually expand the scope of robot function restoration depending on whether each verification stage is passed. In the initial stage, only limited functions (e.g., sensor data collection, low-speed movement, etc.) are activated, and upon passing subsequent stages, all functions, such as main drive units, communication modules, and AI functions, can be restored sequentially. The staged authority restoration method can provide the effect of minimizing the possibility of robot malfunction until potential risks are completely resolved. The restoration details for each stage can be recorded in internal logs and external reporting systems.
[0189] In one embodiment, the safety restart protocol module (270) can perform a test run within a limited range in four stages. In the limited test run stage, the robot can perform basic movements (movement, rotation, stopping, obstacle avoidance, etc.) in a predefined safety zone or a virtual simulation environment. The test run is conducted at low speed, low power, and within a limited range of motion, allowing the stability of the system to be monitored in real time. During the test run, sensor data, drive unit response, communication status, and whether an error occurred can be automatically recorded. The results of the test run are evaluated for each item, and if an abnormal sign is detected, the restart procedure can be stopped immediately. The limited test run stage can minimize the risk of malfunction in the actual environment and provide the effect of verifying final safety before restart.
[0190] In one embodiment, the results of a limited test run can be automatically reported to a registration agency. The test run result data can be converted into a standardized reporting format and automatically transmitted to the registration agency's server via a network. The report data may include the time of the test run, operation items, success / failure status, error codes, sensor logs, etc. The automatic reporting function supports external agencies in verifying the safety of the restart process in real time and proceeding with additional inspections or approval procedures if necessary. The automatic reporting of test run results can provide the effect of enhancing transparency and accountability in the restart process.
[0191] In at least one embodiment of the present disclosure, the safety restart protocol module (270) restarts in a sequence of five stages of phased restoration of operational authority, and may allow full operation to resume only when all stages have been passed. The safety restart protocol module (270) may finally restore operational authority for the full functions of the robot when all verification and test operation stages have been successfully completed. The command to resume full operation is automatically triggered through an internal control system and may be reported to an external registration agency in a synchronized manner. Prior to full operation resumption, additional safety inspections or approval procedures may be required. The condition for full operation resumption may be satisfied only when the inspection results of each stage meet all predefined criteria.
[0192] In at least one embodiment of the present disclosure, full resumption of operation may mean that all functions of the robot are restored to normal, allowing it to return to its previous operating state. Upon full resumption of operation, all systems of the robot, such as sensors, actuators, communication, and AI functions, are activated normally, allowing the robot to return to its operating state prior to the activation of the kill switch. Full resumption of operation may be applied in various operating scenarios, such as the robot's return to normal tasks, resumption of services, and restart of automated operations. The full resumption of operation step may provide the effect of ensuring that only robots that have passed all multi-stage verifications can safely perform their full functions.
[0193] In at least one embodiment of the present disclosure, an international cooperation protocol module (250) may be included to define the coordination of authority between relevant countries, joint activation procedures, determination of jurisdictional priority, and dispute resolution mechanisms when a kill switch is activated for a Physical AI robot (humanoid robot, embodied AI robot, human AI robot, etc.) operating across multiple countries. The international cooperation protocol module (250) may automatically calculate the priority of authority based on the robot's operating location, country of registration, and the nature of the threat situation. The international cooperation protocol module (250) may collect the authority grade of each country's regulatory agency, the robot's registration information, and real-time location data in a standardized data format. Based on the collected information, the international cooperation protocol module (250) may execute the logic for determining jurisdictional priority when a request to activate the kill switch is received. The logic for determining jurisdictional priority may be automatically applied according to various scenarios, such as priority of the country of residence, priority of the country of registration, and joint approval. The international cooperation protocol module (250) may notify each country's regulatory agency of the results of the coordination of authority in real time. The international cooperation protocol module (250) can perform multinational electronic signature and multi-authentication procedures in parallel when a joint activation procedure is required. When a dispute occurs, the international cooperation protocol module (250) can automatically extract kill-switch activation history, authority exercise history, and approval procedure data recorded on the blockchain and submit them to a dispute resolution body.
[0194] In one embodiment, the international cooperation protocol module (250) can define the kill-switch authority relationship between the robot's country of origin and the country of registration when the robot crosses a border and is operating in another country. The international cooperation protocol module (250) can obtain real-time location information of the robot through a satellite-based location tracking system. The international cooperation protocol module (250) can automatically determine the time when the robot has entered the country of origin by comparing the obtained location information with the robot's country of registration information. The international cooperation protocol module (250) can transmit authority request signals to the regulatory authorities of the country of origin and the country of registration, respectively. The international cooperation protocol module (250) can grant the regulatory authority of the country of origin the authority to activate the kill-switch up to Level 3 according to the principle of priority of the country of origin. The international cooperation protocol module (250) can notify the regulatory authority of the country of registration of the activation fact and reason in real time. This structure can ensure that there is no gap in kill-switch authority even when the robot moves internationally.
[0195] In one embodiment, the international cooperation protocol module (250) allows the regulatory authority of the host country to prioritize the activation of a kill switch up to Level 3 in the event of an immediate threat to human safety and subsequently notify the registration country. The international cooperation protocol module (250) can prioritize the processing of an emergency activation request from the regulatory authority of the host country when a threat situation occurs. The international cooperation protocol module (250) can generate and propagate a kill switch command of Level 3 or lower solely through the sole authentication of the regulatory authority of the host country. Immediately after activation, the international cooperation protocol module (250) can automatically transmit information such as the fact of activation, threat code, robot identification number, and time of activation to the regulatory authority of the registration country. The international cooperation protocol module (250) can record the subsequent notification details on a blockchain to transparently verify the legitimacy of the activation and compliance with procedures. This mechanism can simultaneously realize rapid response and international sharing of responsibility in situations where human safety is threatened.
[0196] In one embodiment, the international cooperation protocol module (250) requires joint approval from the regulatory authorities of the host country and the registration country when a kill switch of Level 4 or higher is activated, but may allow activation by the host country alone if the time required for joint approval exceeds a preset time. When a command request of Level 4 or higher is made, the international cooperation protocol module (250) may send a message requesting simultaneous approval to the regulatory authorities of the host country and the registration country. The international cooperation protocol module (250) may collect the electronic signature and multi-authentication results of each agency in parallel. The international cooperation protocol module (250) may measure the time required for joint approval using a timer. If the preset approval waiting time elapses, the kill switch command may be automatically activated with only the sole approval of the regulatory authority of the host country. When activated alone, the international cooperation protocol module (250) may automatically guide the regulatory authority of the registration country to post-notification and objection procedures. This structure can provide the effect of preventing response failures caused by approval delays in high-risk situations.
[0197] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) allows the regulatory authority of the country where the robot is located to have priority activation authority in the event of an immediate threat to human safety, and to review the appropriateness of the measures through consultation among the relevant countries afterward. The international cooperation protocol module (250) can automatically approve the priority activation of the regulatory authority of the country where the robot is located for a command containing a human safety threat code. After activation, the international cooperation protocol module (250) can transmit the details of the measures, the threat situation, and the response results to the relevant national regulatory authority in the form of a standardized report. During the post-consultation stage, the international cooperation protocol module (250) can record the opinions of each country's regulatory authority, additional evidence, and plans for compensation and liability sharing. The international cooperation protocol module (250) can store the results of the consultation on a blockchain so that they can be used as objective evidence in the event of a future dispute. This step-by-step mechanism can simultaneously achieve international trust and responsible kill-switch operation.
[0198] In one embodiment, a periodic test step may be additionally included to periodically verify whether the kill-switch protocol is operating normally. The periodic test step may be automatically scheduled according to a predefined cycle (e.g., 1 day, 1 week, 1 month, etc.) in the system operation management module. The system operation management module may generate a test kill-switch command frame at each test time. The generated test command may be transmitted to a Physical AI robot in actual operation (humanoid robot, embodied AI robot, etc.) or a robot dedicated to testing. The periodic test step verifies in advance whether the system is operating normally, thereby enabling a rapid response in the event of a failure.
[0199] In one embodiment, the regular test is conducted in a simulation mode that does not perform actual blocking, and the propagation path, reception time, and verification results of the test kill-switch command can be recorded. The system operation management module can propagate the test kill-switch command in the simulation mode along the same path as the actual command. Each propagation path (satellite, mobile communication, internet, emergency frequency) can independently receive the test command and record the reception time and path information. A robot or test device can report verification results, such as the digital signature, authority, validity period, and geographical range of the received test command, to the system operation management module. The system operation management module can store the test results in a standardized log format and automatically analyze for any anomalies. Such a simulation test can provide the effect of checking the entire process of propagation, reception, and verification of the system without actual blocking.
[0200] In at least one embodiment of the present disclosure, for a robot in which an abnormality is detected as a result of a test, inspection and repair of the kill-switch system are mandated, and the operation of the robot may be restricted until the inspection is completed. The system operation management module may automatically identify a robot in which abnormal items, such as reception failure, verification error, or response delay, are detected in the test results. For the abnormal robot, an inspection and repair request command may be automatically issued. A robot for which an inspection request has been issued is switched to an operation restriction mode, and its key functions may be restricted until the normal inspection is completed. After the inspection and repair are completed, the system operation management module may re-verify whether normal recovery has been achieved and lift the operation restriction. This procedure may provide the effect of continuously maintaining the reliability and safety of the entire system.
[0201] In one embodiment, a peripheral device linkage safety notification step may be additionally included, which automatically notifies other Physical AI robots (humanoid robots, embodied AI robots, etc.) and IoT devices located near the Physical AI robot of the fact that the kill switch has been activated and the current location of the robot when the kill switch is activated, so that collision avoidance and safety measures can be taken. The peripheral device linkage safety notification step, when the kill switch command is activated,
[0202] In one embodiment, a nearby robot that receives a notification can automatically maintain a safe distance from the robot in question. The nearby robot can calculate a minimum safe distance standard by comparing its own location information with the location coordinates of the activated robot included in the notification message. The nearby robot can automatically adjust its movement path or execute a stop command based on the calculated safe distance. The logic for calculating the safe distance can be dynamically calculated by reflecting the robot's speed, direction, and obstacle information in the working environment. This automatic safe distance maintenance function can provide the effect of minimizing the risk of collision between robots in real time.
[0203] In at least one embodiment of the present disclosure, a surrounding robot that receives a notification may suspend its operation on its own if necessary. The surrounding robot may suspend its operation if the risk level is high by analyzing the blocking level, trigger reason, location information, etc., of the notification message. The logic for suspending the operation may be executed by comprehensively evaluating the robot's current operation status, mission priority, and risk factors in the surrounding environment. After suspending the operation, the surrounding robot may automatically resume operation once a safe state is confirmed. This step-by-step response mechanism can prevent cascading accidents in complex work environments and provide the effect of improving the safety of the entire system.
[0204] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) may define the coordination of authority between relevant countries, joint activation procedures, determination of jurisdictional priority, and dispute resolution mechanisms when a kill switch is activated for a robot operating across multiple countries. The international cooperation protocol module (250) may automatically calculate the priority of authority based on the robot's operating location, country of registration, and the nature of the threat situation. The international cooperation protocol module (250) may collect the authority grades of regulatory agencies in each country, the robot's registration information, and real-time location data in a standardized data format. Based on the collected information, the international cooperation protocol module (250) may execute the logic for determining jurisdictional priority when a request to activate a kill switch is received. The logic for determining jurisdictional priority may be automatically applied according to various scenarios, such as priority of the country of residence, priority of the country of registration, and joint approval. The international cooperation protocol module (250) may notify each country's regulatory agency of the results of the coordination of authority in real time. The international cooperation protocol module (250) can perform multinational electronic signature and multi-factor authentication procedures in parallel when a joint activation procedure is required. When a dispute occurs, the international cooperation protocol module (250) can automatically extract kill-switch activation history, authority exercise history, and approval procedure data recorded on the blockchain and submit them to a dispute resolution body. This mechanism can provide the effect of quickly resolving international disputes, such as conflicts of authority between countries, misuse, and unclear liability.
[0205] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can collect the authority level of each country's regulatory agency, the robot's registration information, and real-time location data in a standardized data format. The international cooperation protocol module (250) can convert the authority level information of each country's regulatory agency into an international standard code and store it. The international cooperation protocol module (250) can query the robot's registration information (registration country, owner, manufacturer, etc.) from a database. The international cooperation protocol module (250) can obtain the robot's real-time location data through a satellite-based location tracking system or a mobile communication network. The international cooperation protocol module (250) can convert the collected data into a standardized data format, such as JSON-LD, and utilize it for internal processing.
[0206] In one embodiment, the international cooperation protocol module (250) can execute jurisdiction priority determination logic when a kill-switch activation request is received based on collected information. When a kill-switch activation request is received, the international cooperation protocol module (250) can compare the robot's real-time location with the country of registration information. The international cooperation protocol module (250) can select the jurisdiction rule to be applied first according to the nature of the threat situation (human life threat, national security, etc.). The international cooperation protocol module (250) can automatically apply one of multiple priority scenarios, such as country of origin priority, country of registration priority, or joint authorization. The international cooperation protocol module (250) can record the results of the jurisdiction priority determination in an internal log.
[0207] In one embodiment, the international cooperation protocol module (250) can notify each national regulatory agency of the authority adjustment results in real time. The international cooperation protocol module (250) can generate the authority adjustment results in a standardized message format. The international cooperation protocol module (250) can transmit the authority adjustment result message to the network address of each national regulatory agency. The international cooperation protocol module (250) can check whether the message transmission was successful and, if it fails, perform a retransmission logic.
[0208] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) may perform multinational electronic signature and multi-authentication procedures in parallel when a joint activation procedure is required. The international cooperation protocol module (250) may simultaneously send electronic signature request messages to each national regulatory authority when a joint activation is required. The international cooperation protocol module (250) may collect electronic signature and multi-authentication information received from each national regulatory authority in parallel. The international cooperation protocol module (250) may verify the collected authentication information and generate a kill-switch command only when approval from all authorities is completed. The international cooperation protocol module (250) may record the authentication and approval history of the joint approval process on a blockchain.
[0209] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can automatically extract kill-switch activation history, authority exercise history, and approval procedure data recorded on the blockchain when a dispute occurs and submit them to the dispute resolution body. The international cooperation protocol module (250) can query kill-switch activation history data from the blockchain network when a dispute occurs. The international cooperation protocol module (250) can extract authority exercise history and approval procedure data from the blockchain. The international cooperation protocol module (250) can convert the extracted data into a standard format for submission to the dispute resolution body. The international cooperation protocol module (250) can automatically transmit the data to the network address of the dispute resolution body.
[0210] In one embodiment, the international cooperation protocol module (250) can define the kill-switch authority relationship between the robot's country of origin and the country of registration when the robot crosses a border and is operating in another country. The international cooperation protocol module (250) can obtain real-time location information of the robot through a satellite-based location tracking system. The international cooperation protocol module (250) can automatically determine the time when the robot has entered the country of origin by comparing the obtained location information with the robot's country of registration information. The international cooperation protocol module (250) can transmit authority request signals to the regulatory authorities of the country of origin and the country of registration, respectively. The international cooperation protocol module (250) can grant the regulatory authority of the country of origin the authority to activate the kill-switch up to Level 3 according to the principle of priority of the country of origin. The international cooperation protocol module (250) can notify the regulatory authority of the country of registration of the activation fact and reason in real time. This structure can ensure that there is no gap in kill-switch authority even when the robot moves internationally.
[0211] In one embodiment, the international cooperation protocol module (250) may grant the regulatory authority of the country of origin the authority to activate the kill switch up to Level 3 according to the country of origin priority principle. The international cooperation protocol module (250) may automatically determine whether the real-time location of the robot is within the country of origin. The international cooperation protocol module (250) may execute logic to grant the authority to activate the kill switch up to Level 3 by verifying the authority level of the regulatory authority of the country of origin. The international cooperation protocol module (250) may notify the regulatory authority of the country of origin of the fact of authorization granting via a standard message.
[0212] In one embodiment, the international cooperation protocol module (250) can notify the regulatory authority of the country of registration of the fact and reason for activation in real time. The international cooperation protocol module (250) can generate a notification message including information such as the fact of kill-switch activation, a code for the reason for activation, a robot identification number, and the time of activation. The international cooperation protocol module (250) can transmit the notification message in real time to the network address of the regulatory authority of the country of registration. The international cooperation protocol module (250) can check whether the notification message has been received and, if it fails, perform retransmission.
[0213] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can ensure that no gap in kill-switch authority occurs even during the international movement of the robot through this structure. The international cooperation protocol module (250) can automatically execute an authority adjustment and notification procedure upon entry into the host country. The international cooperation protocol module (250) can monitor the possibility of an authority gap occurring in real time. If an authority gap is detected, the international cooperation protocol module (250) can immediately send a warning message to the administrator. This mechanism can provide the effect of ensuring the continuity of kill-switch authority in the international movement environment of the robot.
[0214] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) requires joint approval from the regulatory authorities of the country of origin and the country of registration when a kill switch of Level 4 or higher is activated, but may allow activation by the country of origin alone if the time required for joint approval exceeds a preset time. When a request for activation of Level 4 or higher is received, the international cooperation protocol module (250) may simultaneously send an approval request message to the regulatory authorities of the country of origin and the country of registration. The international cooperation protocol module (250) may collect electronic signature and authentication information from each institution in parallel. The international cooperation protocol module (250) may measure the time required for joint approval using a timer. The international cooperation protocol module (250) may automatically allow activation by the country of origin alone if the time required for joint approval (e.g., 30 seconds, 1 minute, etc.) is exceeded. The international cooperation protocol module (250) may immediately notify the country of registration of the fact of activation by the country of origin.
[0215] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can collect and verify electronic signatures and multi-authentication information from each national regulatory body in parallel during a joint approval process. The international cooperation protocol module (250) can simultaneously receive electronic signature data from each national regulatory body. The international cooperation protocol module (250) can collect multi-authentication information (biometric authentication, hardware token, TOTP, etc.) in parallel. The international cooperation protocol module (250) can verify the collected authentication information and generate a kill-switch command only if the authentication of all agencies is valid.
[0216] In one embodiment, the international cooperation protocol module (250) can ensure transparency by recording all authentication and approval history of the joint approval process on a blockchain. The international cooperation protocol module (250) can record all authentication and approval data generated during the joint approval process on a blockchain network along with timestamps. The international cooperation protocol module (250) can ensure the transparency of the approval process and the possibility of post-verification through blockchain records.
[0217] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) may notify the registration country afterward, with the regulatory authority of the country of origin having priority activation authority for kill switches up to Level 3 in the event of an immediate threat to human safety. When the international cooperation protocol module (250) detects a situation threatening human safety, it may prioritize processing an emergency activation request from the regulatory authority of the country of origin. The international cooperation protocol module (250) may generate and propagate a kill switch command of Level 3 or lower solely through the sole authentication of the regulatory authority of the country of origin. Immediately after activation, the international cooperation protocol module (250) may automatically transmit information such as the fact of activation, threat code, robot identification number, and time of activation to the regulatory authority of the registration country. The international cooperation protocol module (250) may record the details of the subsequent notification on a blockchain to transparently verify the legitimacy of the activation and compliance with procedures.
[0218] In one embodiment, the international cooperation protocol module (250) can prioritize processing an emergency activation request from a regulatory agency in the host country when a threat situation occurs. The international cooperation protocol module (250) can detect signals of a threat situation in real time. The international cooperation protocol module (250) can apply logic to process emergency activation requests from regulatory agencies in the host country with higher priority than general requests. The international cooperation protocol module (250) can generate a kill-switch command immediately upon receiving an emergency activation request.
[0219] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can generate and propagate a command for a kill-switch command of Level 3 or lower with only the sole authentication of the regulatory authority of the country of origin. For a command of Level 3 or lower, the international cooperation protocol module (250) can set only the authentication information of the regulatory authority of the country of origin as the verification target. Once the authentication verification is completed, the international cooperation protocol module (250) can generate the command immediately and propagate it through a multipath network.
[0220] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) may automatically transmit information such as the fact of activation, threat code, robot identification number, and activation time to the regulatory authority of the registration country immediately after activation. The international cooperation protocol module (250) may automatically generate a notification message to the regulatory authority of the registration country immediately after the kill-switch is activated. The international cooperation protocol module (250) may include detailed information such as the threat code, robot identification number, and activation time in the notification message. The international cooperation protocol module (250) may transmit the message in real time to the network address of the regulatory authority of the registration country.
[0221] In one embodiment, the international cooperation protocol module (250) can record the post-notification details on a blockchain to transparently verify the legitimacy of the activation and compliance with procedures. The international cooperation protocol module (250) can record the post-notification details on a blockchain network along with a timestamp. The international cooperation protocol module (250) can verify the legitimacy of the activation and compliance with procedures retrospectively through the blockchain record.
[0222] In one embodiment, the international cooperation protocol module (250) can simultaneously achieve rapid response and international responsibility sharing in situations where human safety is threatened. When the international cooperation protocol module (250) detects a signal of threat to human safety, it can immediately execute the independent activation logic of the regulatory agency of the host country. The international cooperation protocol module (250) can record the activation history and details of responsibility sharing on a blockchain and use them as a basis for international responsibility sharing.
[0223] In one embodiment, the international cooperation protocol module (250) may be designed to enable rapid unilateral activation by the regulatory authority of the host country in an emergency situation. The international cooperation protocol module (250) may minimize activation delays by minimizing authentication procedures upon detection of an emergency situation. After unilateral activation, the international cooperation protocol module (250) may simultaneously perform automatic notification and recording procedures to relevant agencies.
[0224] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can record post-notification details and activation history on a blockchain to transparently verify the legitimacy of the activation and compliance with procedures. The international cooperation protocol module (250) can record post-notification details, activation history, and authentication records on a blockchain network along with timestamps. The international cooperation protocol module (250) can objectively verify the legitimacy of the activation and compliance with procedures through the blockchain records.
[0225] In one embodiment, the international cooperation protocol module (250) may perform relief procedures including restricting authority, calculating compensation for the owner of the damaged robot, and taking measures to prevent recurrence, if the activation is determined to be improper based on the results of the post-activation appropriateness review. The international cooperation protocol module (250) may automatically analyze the results of the post-activation appropriateness review. If the activation is determined to be improper, the international cooperation protocol module (250) may execute logic to restrict the authority of the issuer. The international cooperation protocol module (250) may automatically initiate a procedure to calculate compensation for the owner of the damaged robot. The international cooperation protocol module (250) may send a message to relevant agencies recommending measures to prevent recurrence (education, system improvement, etc.).
[0226] In at least one embodiment of the present disclosure, the international cooperation protocol module (250) can automatically extract kill-switch activation history, authority exercise history, and approval procedure data recorded on the blockchain when a dispute occurs and submit them to a dispute resolution body. The international cooperation protocol module (250) can automatically extract kill-switch activation history, authority exercise history, and approval procedure data from the blockchain network when a dispute occurs. The international cooperation protocol module (250) can convert the extracted data into a standard format for submission to a dispute resolution body. The international cooperation protocol module (250) can automatically transmit the data to the network address of the dispute resolution body.
[0227] FIG. 4 shows an example of an electronic device according to one embodiment.
[0228] Referring to FIG. 4, the electronic device (110) may include memory (410) and a processor (430).
[0229] The memory (410) can store instructions (e.g., programs) executable by the processor (430). For example, the instructions may include instructions for executing the operation of the processor (430) and / or the operation of each component of the processor (430).
[0230] The memory (410) can be implemented as a volatile memory device or a non-volatile memory device.
[0231] Volatile memory devices can be implemented as DRAM (dynamic random access memory), SRAM (static random access memory), T-RAM (thyristor RAM), Z-RAM (zero capacitor RAM), or TTRAM (Twin Transistor RAM).
[0232] Non-volatile memory devices can be implemented as EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, MRAM (Magnetic RAM), Spin-Transfer Torque (STT)-MRAM, Conductive Bridging RAM (CBRAM), FeRAM (Ferroelectric RAM), PRAM (Phase change RAM), Resistive RAM (RRAM), Nanotube RRAM, Polymer RAM (PoRAM), Nano Floating Gate Memory (NFGM), holographic memory, Molecular Electronic Memory Device, or Insulator Resistance Change Memory.
[0233] The processor (430) can process data stored in memory (410). The processor (430) can execute computer-readable code (e.g., software) stored in memory (410) and instructions triggered by the processor (430).
[0234] The processor (430) may be a data processing device implemented in hardware having a circuit having a physical structure for executing desired operations. For example, the desired operations may include code or instructions included in a program.
[0235] For example, a data processing device implemented in hardware may include a microprocessor, a central processing unit, a processor core, a multi-core processor, a multiprocessor, an Application-Specific Integrated Circuit (ASIC), and a Field Programmable Gate Array (FPGA).
[0236] The processor (430) can cause the electronic device (110) to perform one or more operations by executing code and / or instructions stored in memory (410). The operations performed by the electronic device (110) have been described in detail with reference to FIGS. 1 to 3, so redundant descriptions will be omitted below.
[0238] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based thereon. For example, appropriate results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents. Therefore, other implementations, other embodiments, and equivalents to the claims below also fall within the scope of the claims.
Claims
Claim 1 In the AI robot global kill-switch standard protocol system, an electronic device; The system includes an AI robot that communicates with the electronic device through a network comprising multiple independent communication paths, wherein the electronic device generates a kill-switch command including a kill-switch command frame in a standardized format that guarantees machine-to-machine interoperability independently of the manufacturer and operating system, based on the unique identification number (ID) of the AI robot; the kill-switch command frame includes a timestamp, a random number (Nonce) uniquely generated for each command, and a sequence number indicating the transmission order of the command; the entire kill-switch command frame is encrypted using a public key-based encryption algorithm to apply end-to-end encryption that is not decrypted at intermediate nodes; and the AI robot generates a propagation signal including the generated end-to-end encrypted kill-switch command and transmits it through each of the multiple communication paths; the AI robot receives the transmitted kill-switch command from the electronic device, decrypts it with a corresponding secret key, and verifies the timestamp, the random number, and the sequence number included in the kill-switch command frame to determine whether it is a repeated reception of the same command or a replay attack that has arrived through the multiple communication paths. Determining, and executing a forced shutdown function of the AI robot based on shutdown level information included in the kill-switch command frame of the received kill-switch command, wherein the shutdown level information designates one of a plurality of shutdown levels having different shutdown intensities, and the execution of the forced shutdown function includes sequentially executing a plurality of operations forming a safety stop sequence to minimize secondary damage caused by sudden shutdown based on the current state information of the AI robot identified based on self-sensor data acquired from a sensor mounted in the AI robot and the shutdown level information, and verifying the execution result of each operation with the self-sensor data, and the AI robotA system that switches to a cutoff level among the plurality of cutoff levels that designates energy supply cutoff when the above safety stop sequence is not completed within a preset maximum allowable time for the above cutoff level information. Claim 2 In claim 1, the kill-switch command frame includes an identification number of the AI robot, a blocking level, a blocking reason code, administrator identification information, a validity period, and a geographical area to which the kill-switch command applies; the plurality of blocking levels includes Level 1, which specifies partial restriction of functions; Level 2, which specifies deactivation of autonomous functions and switching to manual control; Level 3, which specifies suspension of all driving functions and switching to safe mode; Level 4, which specifies energy supply cutoff and external communication cutoff; and Level 5, which specifies locking of physical driving parts; the blocking reason code is a reason value selected from a predefined code table; the validity period includes a start time and an end time for the application of the kill-switch command; and the electronic device generates transmission data encoded in a semantic data structure based on the kill-switch command frame, a system. Claim 3 In paragraph 2, the electronic device generates an authorization verification result applying multi-factor authentication based on the blocking level information, provides an emergency bypass path that allows authentication of a single authorized person based on the input of a pre-set human threat condition, requires ratification by a regular authentication procedure within 24 hours after activation for a kill-switch command activated through the emergency bypass path, and if ratification is not made within 24 hours, automatically releases the kill-switch command and initiates an abuse of authority review against the issuer, wherein the multi-factor authentication is performed by combining multiple authentication means among biometric authentication, hardware tokens, and TOTP according to a pre-defined authorization class for each blocking level, and the authentication through the emergency bypass path includes verifying the issuer's biometric information and location information together. Claim 4 In claim 1, the network comprises a global multipath propagation network comprising a plurality of communication paths including satellite communication, terrestrial mobile communication, the Internet, and a dedicated emergency frequency band, wherein the satellite communication comprises an LEO satellite mesh network, and the dedicated emergency frequency band comprises a UHF band and a VHF band, and the electronic device generates a multi-propagation signal utilizing the plurality of communication paths simultaneously based on the kill-switch command, monitors the propagation status of each of the plurality of communication paths, activates an alternative node for the communication path where a propagation failure is detected, and performs retransmission of the kill-switch command up to three times until propagation success is confirmed, and if a propagation failure is detected in all of the plurality of communication paths, transmits a physical intervention request signal including the location, identification number, and reason for propagation failure of the AI robot to the nearest control center, and the kill-switch command frame of the retransmitted kill-switch command comprises the timestamp, the random number, and the sequence number. Claim 5 delete Claim 6 In claim 1, the AI robot includes a hardware-based dedicated security chip that operates independently of the main operating system of the AI robot, and the kill-switch receiving and execution firmware within the dedicated security chip includes an independent communication receiving unit, a command verification unit, and an operating unit control interface; the kill-switch receiving and execution firmware within the dedicated security chip independently executes the forced blocking function regardless of unauthorized firmware modification and the state of the operating system based on the received kill-switch command; the dedicated security chip verifies the digital signature of the kill-switch receiving and execution firmware through a built-in security bootloader whenever power is applied, and blocks the execution of firmware that fails to verify the digital signature; the dedicated security chip automatically activates the forced blocking function when an attempt to unauthorized modify the kill-switch receiving and execution firmware is detected; and the update of the kill-switch receiving and execution firmware is permitted only when the joint electronic signature of the manufacturer and the registration authority is verified; and the command verification unit A system that independently performs quadruple verification of the digital signature, issuer authority, command validity period, and geographical scope of a kill-switch command. Claim 7 In claim 1, the electronic device generates a blockchain record transaction combined with a timestamp and an electronic signature based on the reason for activation of the kill-switch, administrator identity information, a command propagation path, and execution result data of the AI robot, wherein the command propagation path includes identification information of the communication path through which the kill-switch command was propagated among the plurality of communication paths, a unique identifier of each relay node, and a command arrival time, and the execution result data includes the time of completion of the stop of the AI robot, the stop state, and whether the safety stop sequence was executed. Claim 8 A system according to claim 1, wherein the AI robot, upon receiving a kill-switch release command, generates a multi-stage safety verification result including hardware integrity self-diagnosis and software normal state verification, AI model normal state verification, sensor calibration and environment recognition function verification, and test operation within a limited range, and based on the multi-stage safety verification result being in a passed state, restores the operational authority of the AI robot in stages in the order of limited function restoration, test operation, and full operation resumption. Claim 9 A system according to claim 1, wherein the electronic device, upon activation of a kill-switch for the AI robot operating across multiple countries, automatically calculates the priority of jurisdiction among the relevant countries based on the operating location of the AI robot, the country of registration, and the nature of the threat situation, and performs a joint activation procedure based on the calculated priority of jurisdiction, wherein the priority of jurisdiction is automatically selected from the country of residence priority, the country of registration priority, and joint authorization according to the nature of the threat situation, and wherein the joint activation procedure includes performing multinational electronic signature and multi-authentication procedures in parallel.
Citation Information
Patent Citations
Automatic storage unified distribution system for refrigeration house
CN120833116A
Robot control method and system, robot and electronic equipment
CN121315937A
Operation device and operation monitoring system
JP2024148594A
5G smart factory replay attack detection method and apparatus
KR1020230084832A
Method, system and non-transitory computer-readable recording medium for protecting user from humanoid robot
KR102915951B1
