A general intelligent hub device and control method for a heterogeneous robot integrating sensing and algorithm control

Abstract

The application relates to the technical field of robots, and discloses a general intelligent core device of a heterogeneous robot with integrated sensing, algorithm control and a control method. The device comprises a base adapting module, a shell body assembly, a top layer sensing module, an independent network module and a core computing core. The shell body assembly is arranged on the base adapting module, the top layer sensing module is arranged on the top of the shell body assembly through an elastic connecting assembly, the independent network module is arranged on the outer side wall of the shell body assembly, the shell body assembly has a mounting cavity, the core computing core is arranged in the mounting cavity, and the independent network module and the core computing core are physically isolated through the shell body assembly. The core computing core is in communication connection with the top layer sensing module and the independent network module through internal Ethernet interfaces. The application can improve sensing stability and ensure that the core computing unit does not reduce frequency and is not disturbed during large model reasoning, thereby improving the computing processing effect of a robot side end integrated device.

Description

Technical Field

[0001] This application relates to the field of robotics technology, and in particular to a general intelligent central device and control method for heterogeneous robots that integrates sensing, computing and control. Background Technology

[0002] With the popularization of robotics technology, especially the application of heterogeneous platforms such as quadruped robots (robot dogs), wheeled robots, and drones, higher demands are placed on the versatility of side-end computing and sensing devices mounted on the robot body. However, existing robot side-end integrated devices mainly suffer from the following systemic defects: The perception system suffers from high mechanical and electrical coupling and poor dynamic stability. Existing devices typically rigidly mount high-precision sensors such as LiDAR and cameras directly to the device casing, lacking physical vibration damping design. This results in high-frequency vibrations during the movement of heterogeneous robots, directly causing SLAM mapping drift and visual blurring. Furthermore, the power supply for the perception module often relies directly on the motherboard output, lacking independent voltage regulation and isolation topology. Under high load conditions on the computing unit, voltage fluctuations or cable stress can easily cause sensor power loss or communication abnormalities.

[0003] Tightly coupled network structures lead to severe electromagnetic interference, limited heat dissipation, and chaotic topology. In pursuit of miniaturization, current technologies often encapsulate high-power wireless image transmission / communication modules and the core computing motherboard within the same sealed cavity. This structure easily causes heat buildup, leading to frequency throttling of the core computing platform during large model inference; furthermore, high-frequency communication signals readily generate near-field electromagnetic interference to delicate computing circuits. The lack of standardized cable routing channels and fixing constraints between internal functional boards exacerbates the risk of short circuits and disconnections under long-term vibration.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main purpose of this application is to provide a general intelligent central device and control method for heterogeneous robots that integrates sensing, computing and control. It aims to solve the problems in the prior art where the sensors are directly and rigidly fixed to the shell, high-frequency vibrations are directly transmitted to the sensors, and heat accumulates in the communication module and the core computing motherboard encapsulated in the same sealed cavity, resulting in poor computing and processing performance of the robot side-end integrated device.

[0006] The first aspect of this application provides a heterogeneous robot universal intelligent hub device integrating sensing, computing and control, for installation on the robot body. The heterogeneous robot universal intelligent hub device integrating sensing, computing and control includes a base adapter module, an outer shell assembly, a top-level sensing module, an independent network module and a core computing hub. The outer shell assembly is mounted on the base adapter module. The top-level sensing module is mounted on the top of the outer shell assembly via an elastic connection component to attenuate high-frequency vibrations transmitted from the robot body to the top-level sensing module. The independent network module is mounted on the outer wall of the outer shell assembly. The outer shell assembly has a mounting cavity, and the core computing center is located within the mounting cavity. The independent network module and the core computing center are physically isolated from each other by the outer shell assembly. The core computing center communicates with both the top-level sensing module and the independent network module via an internal Ethernet interface. The top-level perception module is used to collect multimodal raw data around the robot body and send it to the core computing center; The core computing center is used to process the multimodal raw data to generate control commands for the robot body and transmit them to the independent network module. The independent network module is used to send the control commands to the robot body via wireless communication, and / or receive external commands and transmit them to the core computing center.

[0007] Optionally, in one embodiment of this application, the outer casing assembly is provided with a voltage regulator module, which is connected to the core computing hub through a first link and to the top-level sensing module through a second link; The transformer and voltage regulator module is used to receive the power input from the robot body and output independent operating voltages to the core computing center and the top-level perception module, respectively.

[0008] Optionally, in one embodiment of this application, the elastic connection component includes a plurality of damping shock absorbers, which are disposed at the four corners of the bottom end of the top sensing module. The top sensing module is suspended on the top of the outer shell component through the plurality of damping shock absorbers to attenuate the high-frequency vibrations transmitted by the robot body.

[0009] Optionally, in one embodiment of this application, the top-level sensing module includes an inclined mounting platform, a lidar, an RTK antenna module, a vision mounting position, and an integrated camera; The angle between the inclined mounting platform and the horizontal plane ranges from 17 degrees to 23 degrees. The lidar is mounted on the inclined mounting platform. The RTK antenna module is located above the inclined mounting platform. The visual mounting positions are located on the lower sides of the inclined mounting platform. The visual mounting positions are concave inward. The integrated camera is mounted on the visual mounting positions. The lidar is used to collect obstacle data around the robot body, the RTK antenna module is used to determine the positioning data of the robot body, and the integrated camera is used to collect visual data around the robot body.

[0010] Optionally, in one embodiment of this application, the independent network module includes an independent network housing, a cover plate, an adapter plate, and a communication core board. The independent network housing is bolted to the outer wall of the housing assembly. The cover plate is fixedly connected to the independent network housing. The adapter plate is fixedly connected inside the independent network housing. The communication core board is inserted into the adapter plate.

[0011] Optionally, in one embodiment of this application, the core computing hub includes a shock-resistant support frame, a through-support column, and multiple heterogeneous computing boards. The shock-resistant support frame is fixed to the outer shell assembly, and the multiple heterogeneous computing boards are installed on the shock-resistant support frame. The through-support column passes vertically through the fixing holes on the heterogeneous computing boards, so that the multiple heterogeneous computing boards are stacked and locked with the shock-resistant support frame at intervals.

[0012] Optionally, in one embodiment of this application, the outer casing assembly includes a main casing, a support position, an assembly platform, and a heat dissipation grille. The support position and the assembly platform are integrally formed and disposed within the main casing. The heat dissipation grille is disposed on the side wall of the main casing and is arranged in an array, inclined downward along the side wall. The wall surface of the main casing is provided with a light guide groove structure for allowing internal LED light to pass through. The side wall or front of the main casing is provided with dustproof and sound-permeable holes, which are correspondingly disposed with internal speaker and microphone modules.

[0013] Optionally, in one embodiment of this application, the base adapter module includes a fixed base, a CNC metal processing adapter plate, a strip groove, and a wiring channel. The CNC metal processing adapter plate is embedded inside the fixed base. The strip groove is disposed on the bottom surface of the fixed base. The fixed base is connected to the robot body through the strip groove. The wiring channel is disposed inside the fixed base for allowing the main power supply to pass through the body housing from the outside.

[0014] A second aspect of this application also provides a control method for a heterogeneous robot universal intelligent central device integrating sensing and computing control based on any of the above-described solutions, wherein the control method includes: The top-level perception module collects multimodal raw data around the robot body and sends it to the core computing center; The core computing center processes the multimodal raw data to generate control commands for the robot body and transmits them to the independent network module. The independent network module sends the control commands to the robot body via wireless communication, and / or receives external commands and transmits them to the core computing center.

[0015] Optionally, in one embodiment of this application, the multimodal raw data includes obstacle data, positioning data, and visual data, and the control commands include joint position commands and velocity commands; The process of processing the multimodal raw data to generate control commands for the robot body specifically includes: Feature extraction is performed on the obstacle data, the localization data, and the visual data to obtain voxel features, current pose nodes, and semantic features; The voxel features, the current pose node, and the semantic features are input into the edge multimodal large model for cross-modal fusion and inference, and the expected trajectory is output. The desired trajectory is converted into joint position commands and velocity commands.

[0016] Beneficial effects: This application provides a general intelligent central device and control method for heterogeneous robots that integrates sensing, computing and control. Through mechanical decoupling design, this application eliminates the influence of robot body vibration on precision sensors, improves perception stability, and ensures that the core computing unit does not reduce frequency or interfere with large model inference through thermal, electrical and magnetic triple isolation, thus ensuring normal operation of computing power. Through independent voltage regulation power supply and standardized topology constraints, it prevents sensor power failure and cable loosening, thereby improving the computing and processing effect of the robot side-end integrated device. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a disassembly diagram of a preferred embodiment of the heterogeneous robot general intelligent central device integrating sensing, computing and control according to this application; Figure 2 This is a perspective view of a preferred embodiment of the heterogeneous robot universal intelligent central device integrating sensing, computing and control according to this application; Figure 3 This is a system block diagram of the heterogeneous robot general intelligent central device integrating sensing, computing and control, as described in this application; Figure 4 This is a front view of a preferred embodiment of the heterogeneous robot universal intelligent central device integrating sensing, computing and control according to this application; Figure 5 This is a side view of a preferred embodiment of the heterogeneous robot universal intelligent central device integrating sensing, computing and control according to this application; Figure 6 This is a top view of a preferred embodiment of the heterogeneous robot universal intelligent central device integrating sensing, computing and control according to this application; Figure 7 This is a bottom view of a preferred embodiment of the heterogeneous robot universal intelligent central device integrating sensing and computing control according to this application; Figure 8 This is a perspective view of the sliding connection component and the outer shell component of a preferred embodiment of the heterogeneous robot general intelligent central device integrating sensing and computing control according to this application. Figure 9 This is a perspective view of the main shell in the outer shell assembly of a preferred embodiment of the heterogeneous robot general intelligent central device integrating sensing and computing control according to this application; Figure 10 A perspective view of the independent network shell of a preferred embodiment of the heterogeneous robot universal intelligent central device integrating sensing, computing and control according to this application; Figure 11 This is a flowchart of a preferred embodiment of the control method for the heterogeneous robot general intelligent central device integrating sensing, computing and control of this application.

[0019] Explanation of reference numerals in the attached figures: 10. Base adapter module; 20. Outer shell assembly; 30. Top-level sensing module; 40. Independent network module; 50. Core computing hub; 21. Main shell; 22. Flexible connection assembly; 41. Adapter board; 42. Cover plate; 43. Independent network shell.

[0020] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0021] To make the objectives, technical solutions, and effects of this application clearer and more explicit, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of this application and not all possible implementations. Based on the embodiments in this application, those skilled in the art can obtain other embodiments without creative effort, and these embodiments are also within the protection scope of this application.

[0022] First, let's introduce the terms used in the embodiments of this application: GbE, Gigabit Ethernet; RJ45, Registered Jack 45, is a standardized 8-pin modular network interface; USB, Universal Serial Bus; GPIO, General-Purpose Input / Output; Pogo Pin, a spring-loaded pin connector (a type of blind-mating charging / data contact); CPU, Central Processing Unit GPU, Graphics Processing Unit ARM, Advanced RISC Machines, is a reduced instruction set processor architecture. CUDA, Compute Unified Device Architecture; Ampere, an Ampere architecture GPU; TOPS, Tera Operations Per Second, is a unit for measuring AI computing power. AGX, Jetson AGX series edge computing modules; RK3588, Rockchip RK3588, Rockchip RK3588 chip (an ARM architecture AIoT processor). XT60 / XT30, high-current plug-in power connector specifications (60A / 30A level). SLAM, Simultaneous Localization and Mapping; RTK, Real-Time Kinematic, is a real-time dynamic differential positioning technology. RGBD, Red Green Blue Depth, is a color depth camera (capable of simultaneously acquiring color images and depth information). FOV, Field of View; LiDAR, Light Detection and Ranging. Ubuntu, an open-source Linux operating system based on Debian; Linux (no abbreviation, operating system kernel name), a Unix-like operating system kernel; SDK, Software Development Kit; API, Application Programming Interface; VLA, Vision-Language-Action Model (Embodied Intelligence Large Model Architecture). CNC stands for Computerized Numerical Control, which refers to high-precision metal cutting processes. DC, Direct Current; IP55, Ingress Protection 55, dustproof and waterproof rating (dustproof: limited dustproof; waterproof: water-resistant). LED, Light Emitting Diode; PCB, Printed Circuit Board; P12A is a specification of high-strength nylon engineering plastic.

[0023] The related technologies also have the following shortcomings: the physical form is fixed and the external interface lacks scenario-based constraints, making it difficult to deploy universally across platforms. The base of existing equipment is usually deeply bound to the non-standard holes of specific robot models, and cross-platform migration requires redesigning the shell. Its external electrical interfaces (such as network ports, USB, power) and antenna layout often lack anti-interference design, failing to effectively avoid the risk of collisions directly in front of the robot during operation; moreover, the pure plastic shell base is prone to tolerance accumulation under long-term load deformation, and the pure metal shell cannot guarantee lightweight, increasing the robot load and reducing the overall installation rigidity of the equipment, which cannot meet the reliability requirements of rapid assembly and actual operation of multi-form robots.

[0024] The general-purpose intelligent central device for robots in this application features physical shock absorption and modular universal interfaces in its mechanical architecture, and electromagnetic and thermal isolation, independent voltage regulation power supply, and standardized wiring topology in its electrical architecture.

[0025] The heterogeneous robot general intelligent central device integrating sensing, computing and control in this application adopts a mechanical composite stacking and electrical independent topology fusion design in terms of system architecture.

[0026] In this embodiment, the physical isolation and electromagnetic protection structure is as follows: the independent network module 40 adopts a split packaging design, has an independent shielding shell, and is externally mounted on the side wall of the shell assembly 20. This structure physically isolates the high-frequency communication unit from the core computing hub 50, achieving dual isolation from thermal circuits and near-field electromagnetic interference.

[0027] In this embodiment, a flexible damping and rigid adaptation structure is used: the top-level sensing module 30 is suspended and mounted on top of the outer shell assembly 20 via an elastic connection mechanism (coil spring and damping assembly) to support the LiDAR, vision sensor, and RTK module. The base adapter module 10 is fitted with a CNC metal plate and has a standardized multi-hole / slot mechanical interface to ensure the rigidity and accuracy of the equipment for cross-platform installation.

[0028] In this embodiment, the electrical topology and regulated power supply architecture are as follows: The device integrates a transformer and voltage regulator module, which is responsible for converting the external wide-voltage DC input into multiple stable voltages, which independently power the core computing hub 50 and the top-level sensing module 30. The core computing hub 50 establishes a fixed hard-connection communication topology with the sensing module and the external network module through its internal Ethernet port and data bus.

[0029] In this embodiment, the housing shape and interface constraints are designed as follows: the housing assembly 20 has an inclined heat dissipation grille on its side, and the external wiring ports (network port, USB, power) and antenna interface are uniformly arranged on the rear panel or side panel to avoid physical interference from the front. The housing has a recessed socket positioning slot and cable management path, and embeds an LED status indicator array controlled by GPIO and an acoustic interaction module.

[0030] This application has the following technical effects: Anti-interference performance and heat dissipation efficiency: By externally encapsulating the network module, electromagnetic interference from high-frequency image transmission to the core computing hub 50 during large model inference is effectively eliminated. The side wall heat dissipation grilles and thermal isolation design of the outer casing component 20 prevent heat accumulation and ensure that the computing unit does not throttle under continuous high load operation.

[0031] Stable perception and system operation capabilities: The elastic damping design of the top-level perception module 30 forms a mechanical filtering layer, effectively isolating high-frequency rigid vibrations during the movement of legged / wheeled robots and improving the stability of SLAM mapping. The transformer and regulator module provides system-level electrical stability, solving the problem of sensor power loss or communication interruption caused by voltage fluctuations.

[0032] Highly reliable wiring topology and anti-interference capabilities: The rear-mounted constraint design and recessed positioning groove of the outer casing assembly 20 reduce the risk of frontal physical impacts and cable breakage when the robot operates in complex terrain. The direct internal network port communication topology ensures low latency and high bandwidth for high-concurrency sensor data acquisition.

[0033] High versatility and rapid deployment: Modular external design, wide voltage input compatibility, and CNC high-rigidity base of base adapter module 10 enable the same set of equipment to be adapted to various heterogeneous robots such as quadruped, wheeled chassis and flying robots without the need for customized adapters, significantly reducing the cost of cross-platform secondary development and physical integration.

[0034] The technical solutions of this application will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0035] The preferred embodiment of this application describes a heterogeneous robot universal intelligent central device integrating sensing, computing, and control, such as... Figure 1 , Figure 2 and Figure 3 As shown, the integrated sensing, computing and control heterogeneous robot general intelligent hub device is used to be installed on the robot body. The integrated sensing, computing and control heterogeneous robot general intelligent hub device includes a base adapter module 10, an outer shell component 20, a top-level sensing module 30, an independent network module 40 and a core computing hub 50. The outer shell assembly 20 is mounted on the base adapter module 10. The top-level sensing module 30 is mounted on the top of the outer shell assembly 20 via an elastic connection component 22 to attenuate high-frequency vibrations transmitted from the robot body to the top-level sensing module 30. The independent network module 40 is mounted on the outer wall of the outer shell assembly 20. The outer shell assembly 20 has a mounting cavity, and the core computing center 50 is located within the mounting cavity. The independent network module 40 and the core computing center 50 are physically isolated from each other via the outer shell assembly 20. The core computing center 50 communicates with both the top-level sensing module 30 and the independent network module 40 via an internal Ethernet interface. The top-level perception module 30 is used to collect multimodal raw data around the robot body and send it to the core computing center 50; the core computing center 50 is used to process the multimodal raw data to generate control commands for the robot body and transmit them to the independent network module 40; the independent network module 40 is used to send the control commands to the robot body via wireless communication and / or receive external commands and transmit them to the core computing center 50.

[0036] In one embodiment of this application, the outer shell assembly 20 is provided with a voltage regulator module. The voltage regulator module is connected to the core computing center 50 through a first link and to the top-level perception module 30 through a second link. The voltage regulator module is used to receive the power input of the robot body and output independent operating voltages to the core computing center 50 and the top-level perception module 30 respectively.

[0037] Specifically, see Figures 1-7 The system includes a base adapter module 10, a housing assembly 20 mounted on the base adapter module 10, a top-level sensing module 30 mounted on the top of the housing assembly 20, an independent network module 40 mounted on the side wall of the housing assembly 20, and a core computing hub 50 and a transformer and voltage regulator module mounted inside the housing assembly 20. The independent network module 40 has an independent shielded housing and is mounted on the outer side wall of the housing assembly 20 via connectors, forming a physical isolation from the core computing hub 50. The top-level sensing module 30 is connected to the housing assembly 20 via an elastic connecting component 22, forming a floating and shock-absorbing structure relative to the housing assembly 20. The transformer and voltage regulator module receives external power input and outputs independent operating voltages to the core computing hub 50 and the top-level sensing module 30, respectively. The core computing hub 50 communicates with the top-level sensing module 30 and the independent network module 40 via an internal Ethernet interface.

[0038] The transformer and regulator module receives a wide voltage DC input of 9~36V; the transformer and regulator module is equipped with XT60 and / or XT30 standard power supply interfaces to be compatible with the core computing hub 50 with different power consumption architectures; the transformer and regulator module separates an independent voltage regulation link to power the top-level sensing module 30.

[0039] In this embodiment, the transformer and voltage regulator module is built into the internal cavity or dedicated circuit slot of the main housing 21, and its periphery is designed with a dedicated interface fixing slot for cable positioning and stress relief. Its electrical topology and terminal configuration are as follows: The transformer and voltage regulator module has a high-strength overall power input interface, responsible for receiving a wide-range DC input of 9~36V from an external robot (design margin meets 180W peak power consumption, such as 19V / 9.47A input condition). To address the power consumption differences of different core computing boards, the module is configured with differentiated motherboard power supply terminals (AGX / Orin uses a high-current XT60 interface; RK3588 uses an XT30 interface). This transformer module separates an independent voltage regulation power supply link to provide a stable operating voltage of 12V~22V (further using 15.8V / 3A) for the top-level sensing device. The technical advantages of using the transformer and voltage regulator module in this application are as follows: the differentiated power supply terminal design ensures the safety of current transmission when different computing platforms are performing model inference at full power consumption; the independent voltage regulation and isolation topology ensures that the computing board will not cause the precision sensors to lose power or restart due to bus voltage fluctuations under high load conditions, thus improving the electrical stability of the system; the dedicated interface fixing slot realizes standardized cable management and prevents the risk of short circuits and disconnections caused by vibration.

[0040] In one embodiment of this application, see Figure 1 and Figure 8 The elastic connection component 22 includes multiple damping shock absorbers, which are disposed at the four corners of the bottom end of the top sensing module 30. The top sensing module 30 is suspended on the top of the outer shell component 20 through the multiple damping shock absorbers to attenuate the high-frequency vibrations transmitted by the robot body.

[0041] In one embodiment of this application, the top-level perception module 30 includes an inclined mounting platform, a lidar, an RTK antenna module, a vision mounting position, and an integrated camera; the angle between the inclined mounting platform and the horizontal plane ranges from 17 degrees to 23 degrees, and the lidar is mounted on the inclined mounting platform; the RTK antenna module is disposed above the inclined mounting platform, and the vision mounting positions are disposed on the lower sides of the inclined mounting platform, the vision mounting positions being concave inward, and the integrated camera is mounted on the vision mounting positions; the lidar is used to collect obstacle data around the robot body, the RTK antenna module is used to determine the positioning data of the robot body, and the integrated camera is used to collect visual data around the robot body.

[0042] Specifically, the top surface of the top sensing module 30 is provided with an inclined mounting platform that is lower in the front and higher in the back. The angle between the inclined mounting platform and the horizontal plane is further limited to 20° for mounting a lidar. An RTK antenna module is provided above the inclined mounting platform, and recessed mounting positions are provided on both sides below for integrating a vision camera.

[0043] In this embodiment, the top-level sensing module 30 is located at the very top of the device. Its top surface is designed as an inclined mounting platform, with a lower front and higher rear, and the tilt angle to the ground is further limited to 20°. This platform is used to support the lidar to effectively cover the near-field blind spots of the heterogeneous robot. An integrated RTK antenna module is located above the inclined platform. Embedded mounting positions are located on both sides below the inclined platform for integrating RGBD or fisheye cameras. This assembly is connected to the lower outer shell assembly 20 through elastic connection components 22 (coil springs or equivalent damping shock absorbers) distributed at the four corners. Its electrical topology and data link are as follows: Each high-precision sensor (lidar, camera, RTK) in the sensing integration assembly is powered by an independent regulated link separated from the bottom transformer and regulated module to avoid the voltage fluctuations caused by the high load on the motherboard. In terms of data flow, the sensors in the module establish a high-speed direct data path with the core computing hub 50 through a preset internal sensor network interface (RJ45 Ethernet port) and a dedicated data bus. The technical effects of this application are as follows: The elastic connection component 22 (27) forms a mechanical isolation layer, making the sensing module float, effectively attenuating the high-frequency vibrations generated when the legged or wheeled robot moves, and ensuring the stability of SLAM mapping and visual sensor data from a physical perspective. The independent regulated power supply and the direct network port data link ensure high bandwidth and electrical stability of multimodal concurrent sensing data during acquisition and transmission.

[0044] Furthermore, the flexible connection components 22 are distributed at the four corners, connecting the sensing module and the outer shell component 20. These components include helical springs or damping shock absorbers, allowing the sensing module to be suspended and attenuating high-frequency vibrations transmitted by the robot body, ensuring the stability of SLAM and visual data. The tilted mounting platform, with its lower front and higher rear, allows the LiDAR to tilt downwards for scanning, effectively covering blind spots caused by low obstacles near the robot. The RTK antenna module, located above the tilted mounting platform and fixed to its top, provides centimeter-level absolute positioning information. Embedded vision mounting positions, recessed on both sides below the tilted mounting platform, are used to mount RGBD depth cameras or fisheye cameras, providing some physical protection for the lenses. An internal sensor network interface, located within the module's internal cavity, connects the LiDAR, camera, and other sensor data lines. Through a pre-set RJ45 Ethernet port and data bus, multimodal sensor data is aggregated and directly connected to the core computing hub 50.

[0045] In one embodiment of this application, see Figure 1 and Figure 10The independent network module 40 includes an independent network housing 43, a cover plate 42, an adapter plate 41, and a communication core board. The independent network housing 43 is bolted to the outer wall of the housing assembly 20. The cover plate 42 is fixedly connected to the independent network housing 43. The adapter plate 41 is fixedly connected inside the independent network housing 43. The communication core board is inserted into the adapter plate 41.

[0046] In this embodiment, the independent network module 40 comprises an independent network housing 43 and a cover plate 42, forming a network compartment independent of the host computer. This compartment, as an independent module, is bolted to the outer wall of the housing assembly 20, featuring a backpack-like design. An adapter plate 41 is located inside the housing for securing the antenna and the high-power communication core board. Its electrical topology and data link are as follows: the image transmission or networking communication core board within this module is directly connected to the reserved external network communication interface (1 GbE port) of the core computing hub 50 via a standard Gigabit Ethernet cable led out from the internal adapter plate 41; its power supply is supported by an independent regulated output link allocated by the transformer and voltage regulator module, forming a complete data and power closed loop. The technical advantages of using the independent network module 40 in this application are as follows: The backpack-style external structure creates a shielding barrier in physical space, realizing near-field electromagnetic interference isolation between high-frequency image transmission signals and the core computing hub's 50 large-model inference circuits; at the same time, the heat generated by the network module is directly dissipated to the external environment without passing through the core computing cavity, achieving thermal isolation; the independent network and power supply topology, combined with the external design, greatly improves the plug-and-play and rapid disassembly and debugging efficiency of network devices in different frequency bands.

[0047] Furthermore, the independent network housing 43 is bolted to the outer wall of the housing assembly 20, providing an independent metal shielded cavity to separate the high-frequency communication unit from the core computing cavity. The cover plate 42 is fastened or screwed to the housing, sealing the network compartment and facilitating the disassembly and maintenance of internal components. The adapter board 41 is fixed inside the housing, used to secure the communication antenna connector and the high-power communication core board, and to lead the signal out to the core computing hub 50 via a standard gigabit Ethernet cable. The communication core board is located inside the compartment, plugged into the adapter board 41, and performs data transmission and reception using wireless communication protocols such as 5G, Wi-Fi, image transmission, or self-organizing networks.

[0048] In one embodiment of this application, the core computing hub 50 includes a shock-resistant support frame, a through support column, and multiple heterogeneous computing boards. The shock-resistant support frame is fixed to the outer shell assembly 20, and the multiple heterogeneous computing boards are installed on the shock-resistant support frame. The through support column passes vertically through the fixing holes on the heterogeneous computing boards, so that the multiple heterogeneous computing boards are stacked and locked with the shock-resistant support frame at intervals.

[0049] In this embodiment, the core computing hub 50 is an edge computing platform. To achieve high computing power integration and multi-platform compatibility within a limited space, the mounting base and anti-vibration support frame of the platform are compatible with the standard physical parameters of AGX (such as Orin) and RK3588. The core computing board is fixed by stacking and fixing at intervals with a vertically penetrating support column (14) and a customized special anti-vibration bracket. Its electrical topology and computing power allocation: The computing hub is equipped with multiple Ethernet interfaces (10 GbE port × 1, 1 GbE port × 2), which are used for data acquisition of sensing devices (radar, RGBD camera) in the internal network, as well as image transmission and multi-machine networking communication in the external network. In terms of computing power scheduling, the system calls the hardware acceleration module at the bottom layer to handle basic communication and data flow, thereby reducing the occupation of computing resources by the basic functional modules and ensuring that the core computing power is maximized and concentrated on the inference deployment of large models on the edge side without reducing the frequency. For AGX, the system directly reuses the 5G communication link built into its motherboard; for other architectures, network expansion is carried out through the reserved 5G slot. Internal Interface Links: Includes one dedicated RJ45 Ethernet port for connecting to internal top-level sensing devices; and two internal USB ports, one for connecting an audio output device (speaker) and the other for a reserved audio input device (microphone). External Interface Links: A dedicated expansion cable assembly extends from the core motherboard, exposing two RJ45 Ethernet ports (one for communication with the robot chassis and the other for connection to other external image transmission networking modules), at least two USB ports for user expansion, and two high-frequency antenna interfaces (Wi-Fi and Bluetooth). Software System and Interfaces: The core computing unit runs on an Ubuntu 22.04 Linux operating system and reserves standard data reporting interfaces at the business application layer. The system provides complete C++ and Python SDKs, as well as user-friendly API interfaces, providing underlying software support for secondary development and cross-platform porting of the robot. The technical advantages of adopting the core computing hub 50 of this application are as follows: the high-density vertical stacking and shock-resistant brackets ensure the structural integrity of electronic components under violent movement; the hardware acceleration mechanism ensures low-latency processing of high-concurrency, high-bandwidth multimodal sensing data (visual, point cloud, acoustic); and the standardized system environment and SDK interface significantly reduce the secondary development threshold for end users.

[0050] Furthermore, the heterogeneous computing boards are mounted on a shock-resistant support frame, running the operating system and the large-scale edge model. The shock-resistant support frame is fixed to the support position and mounting platform inside the housing assembly 20, providing high rigidity support for the computing boards while absorbing some residual vibration. Vertical through-support columns pass vertically through the mounting holes of the computing boards, locking the boards and shock-resistant brackets in a staggered stack, achieving vertical staggered stacking and fixing of multi-layer boards (such as core boards and expansion boards) to prevent displacement due to vibration. The internal hardware interface set includes: 1 x 10 GbE Ethernet port, 2 x 1 GbE Ethernet ports, USB interface, GPIO interface, and audio interface, connecting to the sensing module, network module, LED array, speaker, and microphone.

[0051] In one embodiment of this application, see Figure 1 and Figure 9 The outer casing assembly 20 includes a main casing 21, a support position, an assembly platform, and a heat dissipation grille. The support position and the assembly platform are integrally formed and disposed within the main casing 21. The heat dissipation grille is disposed on the side wall of the main casing 21 and is arranged in an array and inclined downward along the side wall. The wall surface of the main casing 21 is provided with a light guide groove structure for allowing internal LED light to pass through. The side wall or front of the main casing 21 is provided with dustproof and sound-permeable holes, which are correspondingly arranged with the internal speaker and microphone module.

[0052] Specifically, the housing assembly 20 has an integrally formed support and assembly platform inside. The core computing hub 50 includes a shockproof support frame compatible with heterogeneous computing boards. The computing boards are fixed to the shockproof support frame by vertically penetrating support columns at intervals. The independent network module 40 has an adapter plate 41 inside for fixing the communication antenna and network core board. The adapter plate 41 leads out a network cable and is directly connected to the network interface reserved in the core computing hub 50. The independent network module 40 is detachably fixed to the side wall of the housing assembly 20 as an integral unit. The front panel of the housing assembly 20 has a flat, non-protruding design. All external connection ports and antenna connectors of the device are uniformly arranged on the rear panel or side panel, and each force-bearing interface has a recessed socket positioning groove around its periphery. The side wall of the housing assembly 20 is provided with an array of downwardly inclined heat dissipation grilles; the wall surface of the housing assembly 20 is provided with a light guide groove structure for the LED status array to transmit light, as well as dustproof and sound-permeable holes for the embedding of speaker and microphone modules.

[0053] In this embodiment, the main shell 21 of the outer shell assembly 20 is made of lightweight nylon P12A material, used to provide overall spatial envelopment and physical protection for the core computing hub 50 and the transformer and voltage regulator module. The interior of the shell has integrally formed corresponding support positions and assembly platforms to meet the geometric dimensions of the heterogeneous computing board (AGX / RK3588) and the transformer module, ensuring the structural stability of the high-density component stack. The side walls have arrayed downward-sloping heat dissipation grilles to meet the convection heat dissipation requirements of the high-power modules inside, while also providing splash resistance. External interface constraints and functional component mounting structures are formed: to meet the physical interference prevention requirements of the heterogeneous robot, the front panel of the shell is designed to be flat and without protrusions; external connection ports (including 2 external RJ45, 2 external USB, power supply interface, and antenna interface) are uniformly arranged on the rear or side panels, and each force-bearing interface has a recessed socket positioning groove around its periphery. The housing has partially reserved physical mounting cavities for interactive components, including light guide grooves for the LED status array and dustproof and sound-permeable openings for speaker and microphone modules. Status indication and human-machine interaction: A high-brightness LED indicator array is located on the exterior surface of the housing to visually display the device's power, system, and network operating status to the user. This LED array is electrically connected directly to the 4-pin GPIO interface of the core computing board and is controlled by its underlying logic. The technical advantages of using the housing component 20 of this application are as follows: The integrated internal support and assembly platform improve the mechanical stability of multi-layered electronic device stacking; the strict rear interface constraints and recessed positioning grooves constitute a physical stress relief structure, effectively preventing frontal collisions and cable pull-offs when the robot traverses narrow or complex environments; the light guide and sound-permeable assembly structure achieves reliable physical embedding of external interactive hardware while ensuring the overall IP55 protection level of the housing.

[0054] Furthermore, the main housing 21 is integrally molded from lightweight nylon P12A material, providing IP55-level physical protection for the internal electronic components. The internal support and assembly platform are integrally molded into the inner wall of the housing, used to precisely fix the shock-absorbing bracket and voltage regulator module of the core computing hub 50, ensuring structural stability during high-density stacking. The inclined heat dissipation grille utilizes natural convection to cool the internal high-power modules, while the downward-sloping design provides splash protection. The rear or side-mounted interface panel integrates two external RJ45 ports, two USB ports, a power interface, and an antenna interface. All external interfaces are uniformly positioned at the rear or side, with no protrusions at the front, preventing collisions that could damage interfaces and cables during robot movement. The recessed socket positioning groove is integrally molded with the housing; the plug is partially embedded in the groove after insertion, increasing the stress area and preventing direct damage to the socket solder joints when cables are pulled. The light guide structure allows internal LED lights to shine through, enabling external visualization of the LED status array. Dustproof and sound-permeable vents correspond to the installation positions of the internal speakers and microphones, ensuring sound transmission while preventing dust from entering the housing.

[0055] In one embodiment of this application, the base adapter module 10 includes a fixed base, a CNC metal processing adapter plate 41, a strip groove, and a wiring channel. The CNC metal processing adapter plate 41 is embedded inside the fixed base. The strip groove is disposed on the bottom surface of the fixed base. The fixed base is connected to the robot body through the strip groove. The wiring channel is disposed inside the fixed base for allowing the main power supply to pass through the body housing 21 from the outside.

[0056] Specifically, the base adapter module 10 has a CNC metal processing adapter plate 41 embedded inside. The bottom surface of the base adapter module 10 has standard mounting holes or strip grooves distributed in a matrix, and the interior has a wiring channel and insulation protection layer reserved for the main power line to pass through.

[0057] In this embodiment, the fixed base in the base adapter module 10 is a universal base. To address the issues of stress fatigue and tolerance accumulation in pure nylon material, a customized CNC machined metal adapter plate 41 is embedded inside the base. The bottom surface has standard mounting holes or strip grooves arranged in a matrix. Electrical safety and wiring structure: The base has a pre-reserved wiring channel and insulation layer for the main power cable to pass through, ensuring the physical safety of the bottom power input link under motion conditions. The technical advantages of using the base adapter module 10 of this application are as follows: The high-rigidity CNC metal insert plate ensures the mechanical strength of the device's connection to the robot body under severe vibration conditions. The standardized multi-hole design allows the same device to be quickly and seamlessly locked to quadruped robots, wheeled chassis, or multi-rotor drones without the need for customized adapters, reducing integration and deployment costs.

[0058] Furthermore, the CNC metal adapter plate 41, embedded within the fixed base and fitted into the nylon base body, serves as an intermediate rigid layer to enhance rigidity. This addresses the fatigue deformation and tolerance accumulation issues of pure plastic materials under long-term vibration, ensuring mechanical strength when connected to the robot body. Strip-shaped grooves are arranged in a matrix on the bottom surface of the fixed base for bolting to the robot chassis, enabling seamless and rapid installation with heterogeneous robot chassis such as quadrupeds, wheels, and multi-rotors, without the need for custom adapters. Power cable routing channels and insulation protection layers are located in pre-reserved channels within the base, allowing the main power cable to pass through from the outside into the housing. This standardizes the path of the bottom power cable, preventing cable compression, wear, or short circuits during operation.

[0059] This application presents the energy flow principle of a heterogeneous robot general intelligent central device integrating sensing, computing, and control, featuring hierarchical independent power supply and electrical isolation: Firstly, regarding wide-voltage input and central distribution, the device receives 9~36V DC wide-voltage power (compatible with different robot platform voltage standards) from the robot chassis via the base power interface. This power first enters the transformer and voltage regulator module built into the outer casing.

[0060] Secondly, independent power supply links are generated. After the transformer and regulator module performs voltage regulation, filtering, and branching processing on the input power supply, it no longer shares a single bus. Instead, it separates two physically independent power supply links: Link A (high-current main power supply): provides high-power power to the core computing hub through the XT60 (for NVIDIA Jetson AGX Orin) or XT30 (for Rockchip RK3588) interface, meeting its large model inference requirements with a peak power consumption of up to 180W. Link B (clean regulated power supply): provides an independent 15.8V / 3A regulated output to the top-level sensing module, powering precision sensors such as LiDAR and cameras.

[0061] The electrical isolation mechanism cuts off the transmission path of bus voltage fluctuations to the sensor power supply circuit when the computing board is under high load, eliminating the risk of "one core at full load, the whole system losing power" from the physical level.

[0062] This application presents the data flow principle of a heterogeneous robot general intelligent central device integrating sensing, computing, and control, including layered data acquisition, high-speed direct pass-through, and heterogeneous inference: Perception Layer: High-frequency vibrations generated by the robot's movement are filtered by a levitation structure composed of elastic connecting components, resulting in significant attenuation before being transmitted to the top-level perception module. This allows the LiDAR and camera to operate in a relatively stable inertial frame, ensuring the thickness of the point cloud and the clarity of the visual image.

[0063] Data generation: The RTK module provides centimeter-level positioning, and the RGBD camera and LiDAR scan the environment in real time to generate multimodal raw data streams.

[0064] Transmission Layer: Internal high-speed direct-connect bus. Sensing data does not pass through any intermediate conversion chips or wireless links, but is directly connected to the core computing hub via an internal gigabit Ethernet port (RJ45) and a dedicated data bus in a hard-wired manner. This topology avoids the latency jitter and bandwidth bottlenecks of wireless transmission, ensuring lossless, low-latency concurrent transmission of 200,000 points / second laser point clouds and high-definition video streams.

[0065] Computation Layer: Hardware acceleration and heterogeneous inference. The core computing hub (based on Ubuntu 22.04 system) calls the underlying hardware acceleration module to handle basic data I / O, freeing up CPU / GPU resources. Core computing power is concentrated on edge-side multimodal large model (such as VLA model) inference to achieve local real-time decision-making. The inference results are converted into control commands (optionally sent to the robot's motion chassis via a reserved interface) and into visualized data or voice interaction content.

[0066] This application presents the communication flow principle of a heterogeneous robot general intelligent central device integrating sensing, computing, and control, and its external thermomagnetic isolation and networking: Physical isolation communication: Data processed by the core computing center (such as transmitted images and status information) is transmitted to a side-mounted independent network module via an external gigabit Ethernet port. Because this module has an independent metal shielded shell and is mounted outside the main cavity: Electromagnetic isolation: Strong near-field electromagnetic radiation generated by high-frequency image transmission / 5G signals is shielded outside the core computing cavity, preventing interference with the memory bus and CPU operations on the motherboard. Thermal isolation: Heat generated by the high-power communication module is directly dissipated to the external environment, preventing it from entering the core computing cavity and causing CPU throttling.

[0067] The independent network module supports quick switching between different frequency band communication protocols (5G / self-organizing network / OcuSync), enabling the device to serve as a standardized node, supporting the distributed self-organizing network collaborative operation of up to 100 heterogeneous robots, with a maximum communication distance of 5km.

[0068] The key differences in this application are mainly reflected in three dimensions: mechanical isolation, electrical topology, and physical adaptation. First, a flexible vibration damping and physical image stabilization architecture introduces elastic connection components between the top-level sensing module and the outer shell components, forming a mechanical isolation layer through physical decoupling. This solves the SLAM mapping drift and data distortion problems caused by high-frequency vibrations in heterogeneous robots (especially legged robots). Second, a backpack-style network compartment with dual electromagnetic and thermal isolation removes the independent network module (image transmission / 5G), which is prone to high heat and radio frequency interference, from the core computing cavity. It is designed as an independent side-mounted compartment, with direct internal network cable connection, achieving a unified electromagnetic shielding, thermal isolation, and rapid assembly / disassembly. Third, a heterogeneous computing power compatibility and independent voltage regulation power supply topology integrates a transformer and voltage regulator module. Differentiated terminals (XT60 / XT30) are configured for heterogeneous platforms such as AGX and RK3588 to receive wide-voltage input, and an independent voltage regulation link is separated to power the sensing sensors, solving the current safety and sensor power failure issues during high-load model inference from the hardware level. Fourth, the anti-interference rear interface is combined with a high-rigidity universal base. The outer shell adopts a rear-mounted interface and recessed positioning groove design to eliminate the risk of frontal collision. The bottom is fitted with a custom CNC metal adapter plate and matrix holes to solve the tolerance accumulation problem of pure plastic shell, and achieve seamless adaptation to different robot platforms without adapters.

[0069] It should be noted that the integrated intelligent hub device for sensing and computing in this application has the following possible design changes and alternatives in terms of mechanical architecture, electrical topology, and thermal management: First, multi-degree-of-freedom pose adjustment replaces fixed tilt angle: The mounting platform for the top-level LiDAR can be changed from a fixed 20° forward tilt structure to a mechanically damped pivot or an electrically driven actuator structure with adjustable pitch angle. This allows the sensor's field of view (FOV) to be dynamically adjusted within the range of 0° to 45° to adapt to the blind zone coverage requirements of robots with different chassis heights.

[0070] Equivalent replacement of damping medium: The elastic connection components described in the claims are not limited to helical springs or damping columns. In actual avoidance designs, they can be equivalently replaced by wire rope dampers, silicone balls, polyurethane damping blocks, or active magnetic levitation damping platforms. As long as their physical location is between the sensing module and the core shell and they have mechanical low-pass filtering function, they all fall within the equivalent protection scope of this invention.

[0071] Second, variations in electrical topology and transformer power supply systems: Blind-mating interfaces replace hard-wired connections: The electrical connection between the device base and the robot body can be changed from an external XT60 / XT30 cable to a high-current Pogo Pin array or board-to-board blind-mating connector integrated at the bottom of the base (CNC metal plate). Combined with mechanical grooves, this enables cableless blind-mating deployment where pushing in locks the connection and locking provides immediate power.

[0072] Adding an independent energy storage topology: For heterogeneous robots that cannot provide a continuous and stable bus voltage (such as some small drones), the input of the transformer and regulator module can be connected in parallel or replaced with a standardized hot-swappable lithium battery module (such as a V-port battery holder). This makes the central device logically completely independent of the robot's power supply, becoming a purely passive mounted device with independent endurance capabilities.

[0073] Third, the deformation of the outer shell thermal management system and materials: Active cooling is an equivalent replacement for passive grilles: For special operating environments where the temperature is above 60°C for a long time, the inclined cooling grilles on the side walls can be equipped with waterproof (IP67) DC brushless fans or liquid-cooled circulating cold plates, changing the current natural convection heat management architecture to a forced convection or phase change heat conduction architecture.

[0074] Electromagnetic shielding upgrade for the outer shell material: The main shell can be changed from lightweight nylon P12A material to die-cast aluminum alloy or carbon fiber composite materials, and the inner wall can be sprayed with conductive silver paint. This modification, while increasing the overall weight slightly, improves the device's resistance to near-field electromagnetic interference to industrial grade.

[0075] Fourth, the layout and topology of independent network modules are modified: Adaptive shift of installation location: The backpack-style independent network module is not limited to being mounted on the side wall of the main shell. Depending on the overall center of gravity calibration requirements of the aircraft or wheeled robot, the shielded cabin can be moved to the front of the outer shell or the rear of the top sensing module, as long as the physical isolation barrier between it and the core computing cavity remains unchanged.

[0076] Wireless signal connectivity: Data transmission between the network module and the core computing hub can be effectively replaced by ultra-short-range millimeter-wave communication (such as the 60GHz band) or optoelectronic converter connection, instead of the internal gigabit Ethernet cable (RJ45 hard connection), in order to further pursue the ultimate physical space decoupling and electrical isolation.

[0077] This application focuses on the hardware support capabilities for edge-side multimodal large-scale models. The device's outer casing integrates speakers, microphone modules, and an LED interaction array. Combined with the hardware acceleration module and ample computing power integrated within the core computing hub, the underlying hardware architecture is fully capable of deploying edge-side multimodal large-scale models, natural language large-scale models, and other embodied intelligence models (such as VLA). This design aims to provide robots with localized processing capabilities based on natural voice interaction, real-time spatial modeling, and ontological state feedback, making the device a true embodied intelligence hub.

[0078] This application focuses on expanding swarm intelligence networking for multi-machine heterogeneous collaboration. The external and modular design of the independent network modules in this device is not only for thermal management and electromagnetic isolation considerations, but also to adapt to distributed multi-agent collaboration in complex environments. By quickly plugging and unplugging external network modules with different frequency bands or communication protocols (such as self-organizing networks, 5G, OcuSync, etc.), this device can serve as a standardized physical node in a multi-agent network, supporting algorithm sharing, task collaboration, and cross-domain information flow.

[0079] This application features a development ecosystem that simultaneously decouples hardware and software, with a high degree of mapping between physical modular design (such as independent power supply topologies for different boards for transformer and regulator modules, and data buses with standard gigabit Ethernet ports) and the underlying software architecture. The device runs on the Ubuntu operating system and provides standardized APIs and C++ / Python SDKs, simultaneously decoupling hardware shock protection and adaptation from software drivers and environment configuration. This significantly reduces the barriers to secondary development for industry manufacturers, allowing them to focus on upper-layer business logic and domain model inference, and enabling low-cost parallel migration of a single algorithm across various heterogeneous chassis, including quadrupedal, wheeled, and flying vehicles.

[0080] Based on the above embodiments, this application also provides a control method for a heterogeneous robot universal intelligent central device integrating sensing and computing control based on any of the above solutions, such as... Figure 11 As shown, the control method includes the following steps: In step S10, the top-level perception module collects multimodal raw data around the robot body and sends it to the core computing center.

[0081] In step S20, the core computing center processes the multimodal raw data to generate control commands for the robot body and transmits them to the independent network module. In step S30, the independent network module sends the control commands to the robot body via wireless communication, and / or receives external commands and transmits them to the core computing center.

[0082] In one possible implementation, the multimodal raw data includes obstacle data, localization data, and visual data, and the control commands include joint position commands and velocity commands. Step S30 specifically includes: extracting features from the obstacle data, localization data, and visual data to obtain voxel features, current pose nodes, and semantic features; inputting the voxel features, current pose nodes, and semantic features into an edge-side multimodal large model for cross-modal fusion and inference, outputting a desired trajectory; and converting the desired trajectory into joint position commands and velocity commands.

[0083] The control method provided in this application is applied to the above-mentioned heterogeneous robot general intelligent central device integrating sensing, computing and control, thereby possessing all the beneficial effects of the above-mentioned heterogeneous robot general intelligent central device integrating sensing, computing and control, which will not be elaborated here.

[0084] In the description of this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0085] In the description of this application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0086] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0087] It should be noted that, in this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0088] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0089] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A heterogeneous robot universal intelligent central control device integrating sensing, computing, and control, characterized in that, The heterogeneous robot universal intelligent hub device, which integrates sensing, computing and control, is designed to be installed on the robot body and includes a base adapter module, an outer shell assembly, a top-level sensing module, an independent network module, and a core computing hub. The outer shell assembly is mounted on the base adapter module, and the top-level sensing module is mounted on the top of the outer shell assembly via an elastic connection component to attenuate the high-frequency vibrations transmitted from the robot body to the top-level sensing module. The independent network module is disposed on the outer wall of the housing assembly, the housing assembly has a mounting cavity, the core computing center is disposed in the mounting cavity, and the independent network module and the core computing center are physically isolated from each other through the housing assembly; The core computing hub is connected to the top-level perception module and the independent network module via an internal Ethernet interface. The top-level perception module is used to collect multimodal raw data around the robot body and send it to the core computing center; The core computing center is used to process the multimodal raw data to generate control commands for the robot body and transmit them to the independent network module. The independent network module is used to send the control commands to the robot body via wireless communication, and / or receive external commands and transmit them to the core computing center.

2. The heterogeneous robot universal intelligent central device integrating sensing, computing, and control according to claim 1, characterized in that, The outer casing assembly contains a voltage regulator module, which is connected to the core computing center via a first link and to the top-level sensing module via a second link. The transformer and voltage regulator module is used to receive the power input from the robot body and output independent operating voltages to the core computing center and the top-level perception module, respectively.

3. The heterogeneous robot universal intelligent central device integrating sensing, computing, and control according to claim 2, characterized in that, The elastic connection component includes multiple damping and shock-absorbing columns, which are disposed at the four corners of the bottom end of the top sensing module. The top sensing module is suspended on the top of the outer shell component through the multiple damping and shock-absorbing columns to attenuate the high-frequency vibrations transmitted by the robot body.

4. The heterogeneous robot universal intelligent central device integrating sensing, computing, and control according to claim 3, characterized in that, The top-level perception module includes an inclined mounting platform, a lidar, an RTK antenna module, a vision mounting position, and an integrated camera; The angle between the inclined mounting platform and the horizontal plane ranges from 17 degrees to 23 degrees. The lidar is mounted on the inclined mounting platform. The RTK antenna module is located above the inclined mounting platform. The visual mounting positions are located on the lower sides of the inclined mounting platform. The visual mounting positions are concave inward. The integrated camera is mounted on the visual mounting positions. The lidar is used to collect obstacle data around the robot body, the RTK antenna module is used to determine the positioning data of the robot body, and the integrated camera is used to collect visual data around the robot body.

5. The universal intelligent central device for heterogeneous robots integrating sensing, computing, and control according to claim 2, characterized in that, The independent network module includes an independent network shell, a cover plate, an adapter plate, and a communication core board. The independent network shell is bolted to the outer wall of the shell assembly. The cover plate is fixedly connected to the independent network shell. The adapter plate is fixedly connected inside the independent network shell. The communication core board is inserted into the adapter plate.

6. The heterogeneous robot universal intelligent central device integrating sensing, computing, and control according to claim 2, characterized in that, The core computing hub includes a shock-resistant support frame, a through-support column, and multiple heterogeneous computing boards. The shock-resistant support frame is fixed to the outer shell assembly, and the multiple heterogeneous computing boards are installed on the shock-resistant support frame. The through-support column passes vertically through the fixing holes on the heterogeneous computing boards, so that the multiple heterogeneous computing boards are stacked and locked with the shock-resistant support frame at intervals.

7. The universal intelligent central device for heterogeneous robots integrating sensing, computing, and control according to claim 2, characterized in that, The outer casing assembly includes a main casing, a support position, an assembly platform, and a heat dissipation grille. The support position and the assembly platform are integrally formed and disposed within the main casing. The heat dissipation grille is disposed on the side wall of the main casing and is arranged in an array, tilting downwards along the side wall. The wall surface of the main casing is provided with a light guide groove structure for allowing internal LED light to pass through. The side wall or front of the main casing is provided with dustproof and sound-permeable holes, which are corresponding to the internal speaker and microphone modules.

8. The heterogeneous robot universal intelligent central device integrating sensing, computing, and control according to claim 7, characterized in that, The base adapter module includes a fixed base, a CNC metal processing adapter plate, a strip groove, and a wiring channel. The CNC metal processing adapter plate is embedded inside the fixed base. The strip groove is located on the bottom surface of the fixed base. The fixed base is connected to the robot body through the strip groove. The wiring channel is located inside the fixed base and is used to allow the main power supply to pass through the body shell from the outside.

9. A control method for a heterogeneous robot universal intelligent central device integrating sensing, computing, and control as described in any one of claims 1 to 8, characterized in that, The control method includes: The top-level perception module collects multimodal raw data around the robot body and sends it to the core computing center; The core computing center processes the multimodal raw data to generate control commands for the robot body and transmits them to the independent network module. The independent network module sends the control commands to the robot body via wireless communication, and / or receives external commands and transmits them to the core computing center.

10. The control method according to claim 9, characterized in that, The multimodal raw data includes obstacle data, positioning data, and visual data; the control commands include joint position commands and velocity commands. The process of processing the multimodal raw data to generate control commands for the robot body specifically includes: Feature extraction is performed on the obstacle data, the localization data, and the visual data to obtain voxel features, current pose nodes, and semantic features; The voxel features, the current pose node, and the semantic features are input into the edge multimodal large model for cross-modal fusion and inference, and the expected trajectory is output. The desired trajectory is converted into joint position commands and velocity commands.

Images