Tunnel accident rapid guiding system based on visual guiding robot

Abstract

The invention relates to the technical field of highway tunnel traffic safety and emergency response, in particular to a tunnel accident rapid guiding system based on a visual guiding robot, which comprises a central control platform, a tunnel top rigid sliding rail, a tunnel guiding robot capable of moving along the sliding rail and a positioning and communication infrastructure, and the system is linked with an existing tunnel fire alarm and traffic incident detection system. When an accident occurs, the platform receives alarm information and then generates a character / graph guide instruction, the robot is dispatched to move to an accident area along the sliding rail, guide information is projected to a wall surface or a road surface through high lumen projection, interaction between the platform and field personnel is achieved by means of two-way voice, and meanwhile the field environment is sensed and data is transmitted back. Accident response efficiency can be improved, accurate and reliable guidance is guaranteed, deployment and maintenance cost is reduced, and safe evacuation of personnel in tunnel accidents is effectively guaranteed.

Description

Technical Field

[0001] This invention relates to the field of highway tunnel traffic safety and emergency response technology, specifically to a rapid tunnel accident guidance system based on a visual guidance robot. Background Technology

[0002] As a critical enclosed space in transportation networks, tunnels are particularly vulnerable in emergencies such as fires and traffic accidents. Due to their confined space, obstructed visibility, and complex evacuation routes, rapid and effective guidance is crucial for ensuring personnel safety and minimizing accident losses. However, current tunnel accident guidance technology still has many shortcomings, specifically:

[0003] Existing static signs (such as emergency lights and signs) are easily obstructed by smoke and congestion, have insufficient maximum effective visibility distance, and the information is fixed and cannot be dynamically adjusted as the accident develops, which can easily lead to people not receiving the latest guidance information for a certain period of time, affecting evacuation efficiency.

[0004] Existing tunnel broadcasting systems are susceptible to the effects of the semi-enclosed environment of tunnels, causing broadcast content to overlap and echo, resulting in unclear and unambiguous guidance information. This can lead to confusion among personnel and difficulty in correctly understanding the guidance information, thus affecting evacuation efficiency.

[0005] Traditional manual guidance is slow to respond and cannot proactively guide on-site personnel to carry out emergency response measures in a timely manner. In addition, there are safety risks at the accident site (such as fire, collapse, congestion, and thick smoke), making it difficult to quickly reach the core area to carry out guidance.

[0006] Although some tunnels use robot guidance with positioning information from tunnel monitoring videos, the visual positioning accuracy is insufficient, which can easily lead to guidance deviations. The existing active guidance measures of robots mainly include warning lights, directional arrows, broadcasts, intercoms, and small LED screens, which cannot efficiently and directly attract the attention of on-site personnel, resulting in insufficient actual guidance efficiency.

[0007] The existing guidance system has poor compatibility with the existing fire alarm and traffic incident detection systems in the tunnel, requires manual triggering of the guidance process, and is prone to delaying the best guidance time, making it difficult to meet the needs of rapid accident response.

[0008] Therefore, a rapid guidance system for tunnel accidents based on a visual guidance robot is proposed to address the above problems. Summary of the Invention

[0009] The purpose of this invention is to provide a rapid guidance system for tunnel accidents based on a visual guidance robot, so as to solve the problems mentioned in the background art.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A tunnel accident rapid guidance system based on a visual guidance robot includes a central control platform, a rigid slide rail fixedly installed on the top of the tunnel and extending through the entire length of the tunnel, at least one tunnel guidance robot that can move along the rigid slide rail, and positioning and communication infrastructure within the tunnel.

[0012] The central control platform is used to receive accident alarm information sent by the tunnel monitoring system, generate emergency guidance instructions, and dispatch tunnel guidance robots;

[0013] The tunnel guiding robot includes a drive motor, a power supply system, a wireless communication module, a visualization projection device, a two-way voice intercom module, an environmental perception sensor, and a robot control unit. The drive motor is used to drive the suspension movement mechanism, enabling the tunnel guiding robot to move along a rigid slide rail.

[0014] Visualization projection devices are used to project emergency guidance instructions issued by the central control platform onto the tunnel wall or road surface in the form of text or graphics;

[0015] The two-way voice intercom module is used to establish a full-duplex voice communication link between the central control platform and personnel at the accident site;

[0016] The robot control unit is electrically connected to the drive motor, power supply system, wireless communication module, visualization projection device, two-way voice intercom module and environmental perception sensor, respectively, and is used to receive and execute instructions from the central control platform.

[0017] The tunnel positioning and communication infrastructure provides tunnel-guided robots with accurate indoor positioning information and reliable low-latency communication.

[0018] As a preferred embodiment, the suspension movement mechanism includes a power wheel set, a driven wheel set, and a fall arrest mechanism. The power wheel set is driven by a drive motor and clamps or engages with a rigid slide rail. The fall arrest mechanism automatically locks when it detects a loss of force or overspeed, thus fixing the tunnel guide robot to the rigid slide rail.

[0019] As a preferred option, the rigid slide rail serves as both a power supply bus and a communication bus, continuously supplying power to the tunnel-guided robot through its brushes or inductive power modules, and interacting with the central control platform via bus communication.

[0020] As a preferred option, the central control platform also includes an intelligent scheduling module, which is used to automatically calculate the optimal dispatch plan based on the accident location, the real-time location and status of the tunnel guide robot, and can schedule multiple tunnel guide robots to operate collaboratively on a single rigid slide rail.

[0021] As a preferred solution, the environmental perception sensors include high-definition cameras and infrared thermal imagers. The robot control unit is also equipped with an edge computing module for real-time analysis of the video stream, automatic identification of fire, smoke, or personnel entrapment, and real-time transmission of analysis results.

[0022] As a preferred option, the system is linked with the tunnel's existing automatic fire alarm system and traffic incident detection system to receive alarm signals from the automatic fire alarm system and traffic incident detection system as the trigger source for initiating the guidance process.

[0023] As a preferred option, the positioning and communication infrastructure within the tunnel uses an absolute position encoder based on a rigid slide rail to accurately position the tunnel-guided robot.

[0024] As a preferred option, the tunnel-guided robot adopts a modular functional cabin design. The visualization projection device and two-way voice intercom module can be controlled by electric push rods or servo motors to perform lifting or rotation movements in order to find the optimal projection and sound reception angles.

[0025] As can be seen from the technical solution provided by the present invention above, the tunnel accident rapid guidance system based on a visual guidance robot provided by the present invention has the following beneficial effects:

[0026] Significantly improves accident response efficiency: Multi-source alarm linkage (fire and traffic incident system collaboration) combined with intelligent dispatch can quickly complete the process from accident alarm to guidance activation, which is significantly more efficient than traditional manual guidance and buys critical time for personnel evacuation;

[0027] Ensuring accurate and effective guidance information: Precise positioning combined with high-lumen calibrated projection ensures clear and legible guidance text / graphics. Coupled with two-way voice interaction, it quickly attracts the attention of on-site personnel, proactively guides them in emergency response, and prevents them from getting lost, panicking, fearing, or feeling helpless, effectively improving evacuation efficiency.

[0028] Enhanced system stability: Fall protection lock, robot redundancy backup, and backup communication channel design enable the system to adapt to complex environments such as tunnel smoke and electromagnetic interference, preventing guidance interruptions caused by faults and ensuring reliable system operation;

[0029] Reduced deployment and maintenance costs: Rigid slide rails reuse power supply, communication, and positioning functions, reducing the need for fault repair; they are compatible with existing tunnel alarm systems, eliminating the need for repeated construction, while accurate fault location can shorten maintenance time and reduce long-term costs;

[0030] Efficiently ensure personnel safety: Dynamic guidance (adjusting guidance positions as the accident develops) combined with environmental perception (identifying personnel, fire, and other conditions) ensures that evacuation routes are adapted to the on-site situation in real time, minimizing the risk of casualties in accidents. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the overall structure of a rapid tunnel accident guidance system based on a visual guidance robot according to the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0033] To better understand the above technical solutions, the following will provide a detailed description of the technical solutions in conjunction with the accompanying drawings and specific embodiments.

[0034] like Figure 1 As shown, this embodiment of the invention provides a rapid tunnel accident guidance system based on a visual guidance robot, including a central control platform, a rigid slide rail fixedly installed on the top of the tunnel and extending through the entire length of the tunnel, at least one tunnel guidance robot that can move along the rigid slide rail, and positioning and communication infrastructure within the tunnel.

[0035] The central control platform is used to receive accident alarm information sent by the tunnel monitoring system, generate emergency guidance instructions, and dispatch tunnel guidance robots;

[0036] The tunnel guiding robot includes a drive motor, a power supply system, a wireless communication module, a visualization projection device, a two-way voice intercom module, an environmental perception sensor, and a robot control unit. The drive motor is used to drive the suspension movement mechanism, enabling the tunnel guiding robot to move along a rigid slide rail.

[0037] Visualization projection devices are used to project emergency guidance instructions issued by the central control platform onto the tunnel wall or road surface in the form of text or graphics;

[0038] The two-way voice intercom module is used to establish a full-duplex voice communication link between the central control platform and personnel at the accident site;

[0039] The robot control unit is electrically connected to the drive motor, power supply system, wireless communication module, visualization projection device, two-way voice intercom module and environmental perception sensor, respectively, and is used to receive and execute instructions from the central control platform.

[0040] The tunnel positioning and communication infrastructure provides tunnel-guided robots with accurate indoor positioning information and reliable low-latency communication.

[0041] In this embodiment, the suspension movement mechanism includes a power wheel set, a driven wheel set, and a fall arrest mechanism. The power wheel set is driven by a drive motor and clamps or engages with a rigid slide rail. The fall arrest mechanism automatically locks when it detects loss of force or overspeed, fixing the tunnel guide robot to the rigid slide rail. The rigid slide rail also serves as a power supply bus and a communication bus, providing backup power to the tunnel guide robot through its brushes or inductive power extraction module, and interacting with the central control platform via bus communication.

[0042] In this embodiment, the central control platform also includes an intelligent scheduling module, which is used to automatically calculate the optimal dispatch plan based on the accident location, the real-time location and status of the tunnel guide robot, and can schedule multiple tunnel guide robots to operate collaboratively on a single rigid slide rail.

[0043] Furthermore, the central control platform serves as the "command center" of the tunnel accident rapid guidance system based on visual guidance robots. It coordinates the receipt of accident alarm information, generates precise emergency commands, intelligently dispatches tunnel guidance robots, and integrates with existing tunnel systems to ensure efficient initiation and coordinated execution of the guidance process after an accident. It is the core module ensuring rapid system response and precise guidance. The following will elaborate on this platform from its overall structure to its details:

[0044] I. Overall Function Overview:

[0045] The central control platform is responsible for three core tasks: First, information reception and triggering, which involves receiving alarm signals from the tunnel monitoring system, the existing automatic fire alarm system in the tunnel, and the traffic incident detection system as the starting source for the guidance process; second, instruction generation and issuance, which generates emergency guidance instructions containing text (such as "Accident 500 meters ahead, evacuate along the right arrow") and graphics (such as evacuation direction arrows and safety exit signs) based on the accident type (such as fire, traffic accident) and tunnel structural parameters (such as tunnel length and evacuation exit location); and third, robot scheduling and coordination, which involves obtaining the location and status (such as battery level and projection device status) of the tunnel guidance robots in real time, determining the optimal dispatch plan through intelligent scheduling algorithms, and scheduling multiple robots to operate collaboratively on a single rigid slide rail. At the same time, relying on the positioning and communication infrastructure in the tunnel, it achieves low-latency instruction issuance and real-time feedback of robot status, ensuring the orderly and efficient progress of the guidance process;

[0046] II. Submodule Composition and Functions:

[0047] (a) Multi-source information receiving and linkage triggering unit:

[0048] Multi-system signal reception: Real-time reception of three types of core signal source data—first, accident alarm information sent by the tunnel monitoring system (such as vehicle collisions, personnel lingering images captured by surveillance cameras, and associated location information); second, alarm signals from the existing automatic fire alarm system in the tunnel (such as alarms triggered by excessive smoke concentration or abnormal temperature, including specific alarm area coordinates); and third, signals from the existing traffic event detection system in the tunnel (such as time, location, and event type data for events like vehicle congestion and abnormal parking). The receiving interface adopts standardized communication protocols (such as Modbus and TCP / IP) to ensure compatibility with existing systems from different manufacturers.

[0049] Signal verification and deduplication: Cross-verification and deduplication are performed on received multi-source signals to avoid false or repeated triggering of the guidance process. For example, when the automatic fire alarm system sends a "fire alarm at tunnel K2+300", the tunnel monitoring screen and traffic incident detection system data at that location are retrieved simultaneously. If the monitoring screen shows smoke and the traffic system detects vehicles stopping, it is determined to be a valid alarm. If only a single system sends an alarm and there is no other data to support it, it is marked as "pending verification" and a manual review process is triggered.

[0050] The guidance process trigger logic is as follows: the verified valid alarm signal is used as the trigger source of the guidance process. The preset trigger level is automatically matched according to the alarm type (such as fire alarm triggering "Level 1 Emergency Guidance", ordinary traffic accident triggering "Level 2 Regular Guidance"). Different levels correspond to different instruction generation priorities and robot scheduling strategies (such as Level 1 guidance prioritizes scheduling the 2 robots closest to the accident point, and Level 2 guidance schedules 1 robot).

[0051] (II) Emergency Guidance Instruction Generation Unit:

[0052] Basic data loading and retrieval: Before the command is generated, two types of basic data are automatically loaded: one is the tunnel static parameters (such as the total length of the tunnel, the location of evacuation passages, the number of lanes, and the distribution of curves / slopes), which are stored in the local database to ensure that they can be quickly retrieved when there is no communication dependency; the other is the real-time accident data (such as the accident location, accident type, whether it is accompanied by fire / smoke, and the estimated number of on-site personnel), which are synchronously obtained from the multi-source information receiving unit and serve as the core basis for the command content.

[0053] Instruction content generation: Based on real-time accident data and tunnel static parameters, generate dual-mode guidance information in "text + graphics" mode; text information should be concise and clear, suitable for quick identification by personnel inside the tunnel (e.g., "Vehicles behind the accident point, please reverse to the evacuation exit at K2+000"; "Pedestrians, evacuate in the direction of the green arrow"); graphic information should conform to visual guidance standards (e.g., use red arrows to indicate the direction of avoidance of dangerous areas and green arrows to indicate the evacuation direction, with arrow lines no less than 20cm wide to ensure that vehicle drivers and pedestrians can clearly identify it within 50 meters).

[0054] Command encoding and adaptation: The generated text and graphic guidance information is encoded into a format that the tunnel-guided robot can parse (e.g., graphic information is converted into pixel matrix data that the projector can recognize, and text information is converted into UTF-8 encoded text data). At the same time, the command parameters are adapted according to the robot model (if the system contains robots with different configurations) (e.g., the brightness parameters and projection angle parameters of the high-lumen laser projector, to ensure that the robot can execute the command directly after it is issued without secondary adjustments).

[0055] (III) Intelligent Scheduling Unit:

[0056] Real-time robot status monitoring: Through the positioning and communication infrastructure within the tunnel, the core status data of all tunnel-guided robots are acquired in real time—first, position information (based on the absolute position encoder and track positioning calibration module of the rigid slide rail, the precise coordinates of the robot on the slide rail are obtained, with an error of no more than 1 meter); second, operating status (drive motor condition, whether the visualization projection device is normal, the signal strength of the two-way voice intercom module, and the remaining battery percentage); and third, task status (whether it is idle, executing, or in a fault state). All data is updated at a frequency of 1 second to ensure that scheduling decisions are based on the latest status.

[0057] The optimal dispatch algorithm is executed as follows: Based on the accident location (e.g., at tunnel K2+500), the robot's real-time location and status, the optimal dispatch plan is automatically calculated. The core logic of the algorithm is as follows: First, select robots that are "idle and without faults"; Second, calculate the movement distance (based on the rail coordinate difference) and movement time of the selected robots to the accident location (based on the rated speed of the drive motor, such as 0.5m / s, calculation time = distance / speed); Third, prioritize the selection of robots with "shortest movement time + remaining battery ≥ 50%"; If multiple robots meet the conditions, the final number of robots to be dispatched (one or more) is determined by further considering the accident's impact range (e.g., a fire affecting a 100-meter area requires two robots for zoned guidance).

[0058] Multi-robot collaborative scheduling: When scheduling multiple robots to run on a single rigid slide rail, a "time window scheduling mechanism" is used to avoid robot collisions. Each robot is assigned a dedicated movement time period and slide rail interval. For example, robot A starts from K1+000 and moves to K2+500 within 0-100 seconds (covering the K1+000-K2+500 interval); robot B starts from K3+000 and moves to K2+600 within 50-150 seconds (covering the K2+600-K3+000 interval), ensuring that the two robots do not have overlapping or conflicting operations on the slide rail. At the same time, different guidance tasks are assigned to each robot (e.g., robot A is responsible for guiding the area to the left of the accident point, and robot B is responsible for guiding the area to the right), achieving collaborative coverage.

[0059] (iv) Data interaction and status feedback unit:

[0060] Two-way data interaction with the robot: Through the positioning and communication infrastructure in the tunnel (relying on the communication bus of the rigid slide rail), low-latency data interaction with the tunnel guidance robot is achieved. On the one hand, emergency guidance instructions (text / graphic coded data) and scheduling instructions (moving target location, task assignment) are sent to the robot with a communication latency of no more than 0.5 seconds. On the other hand, real-time data transmitted back by the robot (such as fire / smoke images collected by environmental perception sensors, analysis results of edge computing modules, and working status of projection devices) are received to provide a basis for instruction adjustment and scheduling optimization.

[0061] Data interaction with positioning infrastructure: Receive precise robot position data sent by the positioning and communication infrastructure (absolute position encoder based on rigid slide rail) in the tunnel, compare it with the position information returned by the robot itself, and if the error between the two exceeds 0.5 meters, trigger the position calibration command to perform secondary calibration with the track calibration positioning module to ensure that the robot's positioning accuracy meets the guidance requirements (such as the projected guidance information must correspond precisely to the actual evacuation route).

[0062] Data recording and status feedback: Three types of core data are recorded in real time: alarm data (alarm source, accident type, occurrence time / location); dispatch data (dispatched robot number, movement path, task content); and execution data (robot execution status, guidance command issuance time, and on-site environment analysis results). All data is stored in a local database and retained for no less than one year. At the same time, the real-time operating status (such as "Robot A has arrived at the accident point and started projection guidance" and "Communication link is normal") is fed back to system administrators through a visual interface (such as a dashboard) for easy manual monitoring and intervention.

[0063] III. Key Technology Principles:

[0064] (I) Principle of Multi-Source Information Linkage Triggering:

[0065] Based on the logic of "cross-validation + priority determination", the system achieves precise triggering of the guidance process. First, alarm signals from multiple systems (monitoring, fire alarm, traffic detection) are received through a standardized interface. Data association algorithms (such as signal matching based on location coordinates) are used to associate alarm information from the same source from different systems (such as a fire alarm and a monitoring smoke image at the same location). Second, false alarm signals are eliminated through preset "signal validity determination rules" (such as at least two systems sending alarm signals from the same source, or one high-priority system (such as an automatic fire alarm system) sending an alarm signal + monitoring image corroboration). Finally, the guidance priority is determined according to the type of accident (fire accident priority > traffic accident priority > personnel stranded priority). The higher the priority, the faster the guidance process starts (e.g., the time from receiving the alarm to issuing the instruction in a fire accident is ≤10 seconds).

[0066] Principle of intelligent scheduling algorithm:

[0067] The optimal scheduling of the robot is achieved by employing a "multi-objective optimization + conflict avoidance" algorithm. The multi-objective optimization aims to minimize movement time, power consumption, and guidance coverage, and is implemented through a weighted summation model (such as the objective function). ,in, For the time of travel, To estimate power consumption, To guide coverage, , , The scheduling priority of each robot is calculated using a weighting coefficient (adjusted according to accident priority), and the robot with the highest priority is dispatched first. Conflict avoidance is based on the "sliding rail interval-time two-dimensional allocation" model, which allocates non-overlapping running intervals and times to multiple robots to avoid collisions and ensure collaborative operation efficiency.

[0068] (III) Low-latency data interaction principle:

[0069] Relying on the "rigid slide rail communication bus + data compression" technology, efficient data transmission is achieved. On the one hand, taking advantage of the characteristic of "rigid slide rail also serving as a communication bus" in claim 3, data is transmitted through bus communication (such as CAN bus). Compared with wireless communication, bus communication has lower latency (≤0.5 seconds) and stronger anti-interference ability (the electromagnetic environment inside the tunnel is complex, and bus communication can reduce signal attenuation). On the other hand, the issued guidance instructions (such as graphic data) are compressed (using the JPEG2000 compression algorithm) to reduce the amount of data, further reduce transmission latency, and ensure that the instructions reach the robot quickly.

[0070] (iv) Data recording and feedback principles:

[0071] Based on the "structured database + visualization rendering" technology, secure data storage and intuitive feedback are achieved. Relational databases (such as MySQL) are used to store alarm, scheduling, and execution data in a structured manner, and data query efficiency is improved through index optimization (such as creating indexes by accident time and robot number). At the same time, data visualization technology (such as ECharts) is used to convert real-time operating data into dashboards (such as robot status distribution maps and accident handling progress bars), so that managers can intuitively understand the system's operating status. When an anomaly occurs (such as robot failure), the dashboard automatically highlights an alarm, triggering manual intervention.

[0072] IV. Platform Workflow:

[0073] (a) Initialization phase:

[0074] After the central control platform is started, it completes hardware self-checks (such as communication interfaces, computing modules, and database servers) to ensure that all hardware components are operating normally. At the same time, it checks the communication links with the tunnel monitoring system, automatic fire alarm system, traffic incident detection system, tunnel guidance robot, and positioning infrastructure. If the link is interrupted, it triggers a communication repair prompt (such as "Communication with the fire alarm system is interrupted, please check the interface").

[0075] Loading preset parameters: including tunnel static parameters (length, evacuation exit locations, sliding rail coordinate mapping table), scheduling algorithm parameters (weight coefficients) , , The system includes robot rated speed, linkage triggering rules (signal validity judgment conditions, accident priority), and establishes standardized communication connections with various external systems.

[0076] (II) Information Receiving and Verification Stage:

[0077] It receives alarm signals from the tunnel monitoring system, automatic fire alarm system, and traffic incident detection system in real time, and marks each signal with attributes such as "source, time, location, and event type".

[0078] Cross-verify the received signals: For example, if the automatic fire alarm system sends a message "Fire at K2+300", then retrieve the tunnel monitoring screen at that location (to determine if there is smoke) and the traffic incident detection system data (to determine if there are vehicles stopped). If both sets of data corroborate the fire, then it is determined to be a valid alarm; if only a single signal is not corroborated, then it is marked as "to be verified" and pushed to the management personnel interface for manual confirmation.

[0079] (III) Emergency Instruction Generation Phase:

[0080] Based on the "accident type + location" of the effective alarm, call the tunnel static parameters (e.g., the evacuation exit near K2+300 is the left exit of K2+200) to determine the guidance content: text instructions (e.g., "Fire ahead at K2+300, evacuate to the left exit of K2+200 along the green arrow") and graphic instructions (e.g., a green arrow pointing from the accident point to the evacuation exit, with an arrow line width of 20cm and a length of 50cm).

[0081] The generated text and graphic instructions are encoded: text instructions are converted to UTF-8 encoding, and graphic instructions are converted into pixel matrix data that can be recognized by the robot projection device (resolution adapted to high-lumen laser projectors, such as 1920×1080), ensuring that the robot can directly parse and execute them;

[0082] (iv) Intelligent scheduling and instruction issuance phase:

[0083] Retrieve real-time status data (location, battery level, task status) of all tunnel guidance robots, and use an intelligent scheduling algorithm to select the optimal robot (e.g., robot A, which is idle, has a battery level ≥ 50%, and the shortest travel time to the accident point); if the accident has a large impact range (e.g., more than 100 meters), select two robots (A and B) and assign them to different guidance zones.

[0084] Generate scheduling instructions for the selected robots, including the target location of movement (e.g., robot A to K2+350, robot B to K2+450), task division (A is responsible for guiding the K2+300-K2+350 interval, and B is responsible for guiding the K2+350-K2+400 interval), and guidance instructions (encoded text / graphic data).

[0085] The scheduling and guidance commands are sent to the robot via the rigid slide rail communication bus, and the robot's "command reception confirmation" signal is received at the same time. If no confirmation is received within 10 seconds, the command is resent to ensure that the command is delivered.

[0086] (v) Execution monitoring and data recording phase:

[0087] Receive execution data transmitted back by the robot in real time, including movement progress (e.g., "Robot A has moved to K2+320, 30 meters remaining"), projection status (e.g., "Projection device is normal, guidance information is clear"), and environmental perception results (e.g., "High-definition camera has captured 3 stranded people, infrared thermal imager has not detected high temperature points").

[0088] The scheduling strategy is adjusted based on the returned data: for example, if robot A reports "smoke in the projection area, and the guidance information is blurry", then the instruction "adjust the projection angle to the road surface (the original projection was on the wall)" is issued; if robot B suddenly malfunctions, then the idle robot C is rescheduled to replace it.

[0089] Record all data in real time during alarm, command generation, scheduling, and execution processes, store it in the local database, and update the system's operating status in the visual interface (e.g., "During accident guidance, 2 robots are performing tasks normally").

[0090] (vi) Conclusion:

[0091] Once the incident is resolved (e.g., when management sends an "incident cleared" instruction, or when the robot reports "no personnel remaining on site, guidance complete"), the central control platform sends a "task completed" instruction to all robots. The robots then stop guiding and return to their preset standby positions (e.g., the sliding rail node at the tunnel entrance).

[0092] Stop data acquisition and scheduling operations, close temporary communication links with external systems, generate an "accident guidance report" (including alarm details, scheduling process, execution results, and anomaly handling records), and push it to the administrator's email address;

[0093] Save all operational data from this incident, update system parameters (such as robot power consumption records for subsequent scheduling algorithm optimization), release temporarily occupied computing resources, and restore to standby state.

[0094] In this embodiment, the environmental perception sensor includes a high-definition camera and an infrared thermal imager. The robot control unit is also equipped with an edge computing module for real-time analysis of the video stream, automatic identification of fire, smoke or personnel lingering status, and real-time transmission of analysis results.

[0095] The tunnel guiding robot adopts a modular functional cabin design. The visualization projection device and two-way voice intercom module can be controlled by electric push rods or servo motors to perform lifting or rotating movements in order to find the best projection and sound reception angle.

[0096] Furthermore, the tunnel guidance robot is the "execution terminal" of a rapid tunnel accident guidance system based on a visual guidance robot. It moves along a rigid slide rail, providing precise guidance at the accident site through visual projection and two-way voice communication. It also possesses environmental perception and status feedback capabilities, serving as a crucial link between the central control platform and the accident site. The following section provides a detailed description of this robot, from its overall structure to its specific details:

[0097] I. Overall Function Overview:

[0098] The core function of the tunnel guidance robot is to receive and execute scheduling instructions from the central control platform, quickly move to the accident area along a rigid slide rail on the tunnel ceiling, project guidance information (text / graphics) onto the tunnel wall or road surface through a visualization projection device, and achieve real-time communication between the central platform and on-site personnel through a two-way voice intercom module. In addition, it collects on-site data (such as fire and personnel status) through environmental perception sensors, analyzes it locally, and transmits it back to the central platform to provide a basis for instruction optimization. Its design relies on a suspended moving mechanism to ensure high stability, and its modular design enhances adaptability, enabling it to work continuously and reliably in complex tunnel environments (smoke, low light, electromagnetic interference).

[0099] II. Core Components and Functions:

[0100] (a) Suspension and movement mechanism:

[0101] As the connecting carrier between the robot and the rigid slide rail, it undertakes the functions of motion support and safety assurance, specifically including:

[0102] Power wheel set: Directly driven by a drive motor, it is connected to a rigid slide rail using a "double wheel clamping" or "gear meshing" structure (adapted to the cross-sectional shape of the slide rail, such as I-shaped or C-shaped), providing power for movement along the slide rail, with a rated moving speed of 0.5-1m / s, ensuring rapid arrival at the accident point;

[0103] Driven wheel assembly: Distributed in front of and behind the power wheel assembly, it is tightly attached to the side of the slide rail through a spring preload structure to counteract the swing torque when the robot is suspended, and forms "multi-point contact" with the power wheel assembly to ensure stability during movement (the amplitude of the projected image jitter during movement is ≤ ±2°).

[0104] Fall prevention locking mechanism: Built-in tension sensor and centrifugal brake. When the suspension mechanism loses force (such as a sudden drop in tension of more than 50%, which may cause detachment) or overspeed (exceeding the rated speed by 1.5 times), the brake locks instantly (response time ≤ 0.1 seconds). The robot is fixed by the mechanical engagement of the metal claws and the slide rail to prevent it from falling or sliding out of control.

[0105] (ii) Driving and positioning unit:

[0106] Drive motor: Adopts DC servo motor with stepless speed regulation function (speed can be precisely controlled by commands from the central platform, with an adjustment accuracy of ±0.05m / s), equipped with a gearbox and encoder, providing real-time feedback on the moving distance (accuracy ±0.1m) to ensure accurate arrival at the target position;

[0107] Positioning Assist Module: In conjunction with the positioning infrastructure inside the tunnel, it obtains its precise coordinates (error ≤ 0.5 meters) inside the tunnel by reading the absolute position encoder information on the rigid slide rail, and transmits the position data back to the central control platform in real time, supporting scheduling algorithms to optimize the path;

[0108] (III) Visualization projection device:

[0109] Hardware configuration: It adopts a high-lumen (≥5000 lumens) laser projector, which supports clear imaging in strong light (such as vehicle high beams) or smoke environments. The projection resolution is ≥1920×1080, and the projection distance is adjustable from 1 to 10 meters (adapting to different tunnel cross-section heights).

[0110] Optical path calibration system: Built-in gyroscope and electric focusing module. When moving, the gyroscope detects changes in robot posture (such as tilt angle) in real time and automatically adjusts the projection lens angle (adjustment range ±15°); when briefly stopping, automatic focusing is activated to ensure that the edges of the projected text / graphics are clear (blur ≤5%). For example, when projecting a "green evacuation arrow" at a distance of 5 meters, the arrow line width error is ≤±1cm.

[0111] Projection content adaptation: It can receive encoded graphic / text data from the central platform and supports dynamic updates (such as the arrow flashing frequency changing with the level of urgency: Level 1 emergency guidance flashing frequency 2Hz, Level 2 guidance flashing frequency 1Hz). It can also automatically adjust the contrast according to the tunnel wall material (such as concrete, tile) (adjustment range 50%-100%) to improve readability.

[0112] (iv) Two-way voice intercom module:

[0113] Full-duplex communication: Equipped with a high-sensitivity microphone (pickup distance 3-10 meters, noise immunity ≥80dB) and a waterproof speaker (output power ≥10W, coverage radius 15 meters), supporting real-time two-way communication between the central platform and on-site personnel, with voice delay ≤0.3 seconds and no echo interference;

[0114] Environmental Adaptation: The microphone has a built-in windproof noise reduction algorithm that can filter out airflow noise and vehicle engine noise in the tunnel; the speaker adopts a directional sound design to reduce sound reverberation in the tunnel (reverberation time ≤ 0.5 seconds) and ensure that voice commands are clear and identifiable (such as "Please evacuate to the right").

[0115] (v) Environmental sensing sensors:

[0116] High-definition camera: Equipped with a 20-megapixel visible light camera with a 120° field of view, it supports automatic switching between day and night modes (infrared supplementary light is turned on at night), and can capture accident scene images (such as vehicle collision status and personnel positions), with a video frame rate of 30fps, which is transmitted back to the central platform for remote analysis.

[0117] Infrared thermal imager: resolution 640×512, temperature measurement range -20℃-150℃, can penetrate smoke to identify high temperature points (such as fire sources) or human body outlines (even when obscured by smoke), thermal sensitivity ≤50mK, ensuring accurate detection of personnel lingering or fire spread in low visibility environments.

[0118] Data preprocessing: The video stream acquired by the sensor is first denoised and compressed by the local FPGA chip (using H.265 encoding) to reduce the amount of data transmission (compression ratio 10:1), and then transmitted back through the communication bus to reduce bandwidth usage;

[0119] (vi) Robot control unit:

[0120] Core Processor: Employs an industrial-grade ARM processor (quad-core, 2GHz), integrating an edge computing module to run lightweight AI models (such as YOLOv5s), perform real-time analysis of camera and thermal imager data (processing latency ≤1 second), automatically identify fire (accuracy ≥95%), smoke (accuracy ≥90%), personnel (accuracy ≥98%), and generate structured analysis results (e.g., "Two stranded personnel were found at K2+300, with no obvious high-temperature point").

[0121] Interface and linkage: It is electrically connected to the drive motor, projection device, voice module and sensors through interfaces such as GPIO and RS485. After receiving the instructions from the central platform, it is decomposed into specific execution signals (such as controlling the rotation angle of the motor and the switching on and off of the projection device). At the same time, it summarizes the status data of each module (such as motor current and projection lamp temperature) and forms a "health status report" to be sent back to the platform.

[0122] Local decision backup: When communication with the central platform is interrupted, a preset emergency procedure is automatically started (such as moving at a constant speed along the slide rail and projecting a fixed "evacuate to the exit" command) until communication is restored, ensuring that the guidance function is not interrupted;

[0123] (vii) Energy and communication interfaces:

[0124] Power supply module: DC power (24V / 10A) is obtained from the rigid slide rail (which also serves as the power supply bus) through brushes or inductive power extraction module. A built-in supercapacitor serves as a backup power source (with a runtime of ≥10 minutes) to prevent the robot from losing power when the slide rail power supply is temporarily interrupted.

[0125] Communication interface: It interacts with the central platform through the communication bus of the slide rail (such as the CANopen protocol), with a communication rate of 1Mbps. It supports parallel processing of command issuance, status feedback and video stream transmission, and its anti-interference capability meets industrial-grade standards (can withstand ±2kV electrostatic discharge).

[0126] III. Key Technology Principles:

[0127] (I) High-stability movement principle:

[0128] Stable movement is achieved through a dual guarantee of "mechanical structure + control algorithm": Mechanically, the power wheel set and the driven wheel set form a "three-point positioning" clamping slide rail to counteract the overturning torque generated by gravity; in terms of control, the drive motor adopts PID closed-loop control, which adjusts the output torque in real time according to the speed deviation fed back by the encoder (automatically increasing power when encountering a slight slope of the slide rail), ensuring that the movement speed fluctuation is ≤±0.03m / s, laying the foundation for projection stability;

[0129] (II) Visual Guidance and Precise Projection Principle:

[0130] Projection alignment is achieved based on "spatial coordinate mapping" technology: The robot has a built-in 3D model of the tunnel cross section (pre-stored in the control unit), and calculates the relative positional relationship between the projection lens and the target projection surface (wall / road surface) by combining its own positioning coordinates and gyroscope attitude data. The focal length and angle are dynamically adjusted through the electric focusing module to ensure that the guidance information (such as arrows) is always accurately aligned with the actual evacuation route (such as pointing to the exit), with a deviation of ≤ ±5cm.

[0131] (III) Real-time sensing principle of edge computing:

[0132] The architecture adopts a "lightweight model + hardware acceleration" approach: fire and personnel identification algorithms are deployed on the edge computing module (integrated with an NPU neural network processor) to perform localized analysis of sensor data, avoiding reliance on the computing resources of the central platform; at the same time, through an "incremental learning" mechanism, it regularly receives model update packages (optimized based on historical data) from the platform to improve the recognition accuracy in complex scenarios (such as dense smoke and multiple obstacles).

[0133] (iv) Modular functional cabin design principles:

[0134] Each core component (projection device, voice module) is connected to the robot body through a standardized interface, and "adjustable posture" is achieved through electric push rods or servo motors: for example, the projection device can be raised and lowered vertically (adjustment range 0-50cm) and rotated horizontally (±30°), and the microphone / speaker of the voice module can be turned towards the sound source (driven by the camera to identify the position of the personnel), so as to obtain the best projection visibility and voice interaction effect in different tunnel sections (such as height 3-8 meters) or personnel distribution conditions;

[0135] IV. Robot's Workflow:

[0136] (a) Standby and startup phases:

[0137] The robot is in a default standby position on the slide rail (such as at the tunnel entrance). The control unit performs a periodic self-check (once every 10 seconds): checking whether the power wheel set, projection device, sensors, etc. are normal, and whether the power is sufficient (the voltage is monitored in real time when powered by the slide rail), and sends the "standby normal" status back to the central platform.

[0138] Upon receiving the dispatch instructions (including target location and task type) from the central platform, the control unit parses the instructions, starts the drive motor, unlocks the suspension moving mechanism, and begins to move along the slide rail towards the target location.

[0139] (II) Movement and Positioning Phase:

[0140] During the movement, the drive motor runs at the commanded speed (e.g., 0.8 m / s), the encoder provides real-time feedback on the movement distance, and the positioning auxiliary module synchronously reads the absolute position encoder information of the slide rail and reports the current coordinates (e.g., "K2+350") to the central platform every 0.5 seconds.

[0141] The anti-fall locking mechanism continuously monitors the tension and speed. If any abnormality occurs (such as a sudden increase in resistance due to local deformation of the slide rail), it will immediately trigger deceleration or locking and send an "abnormal alarm" to the platform.

[0142] (III) Guiding Implementation Phase:

[0143] Upon reaching the target location, the control unit stops the drive motor and locks the driven wheel assembly (to prevent slippage). At the same time, the projector's optical path calibration is initiated: the gyroscope detects the body's posture, and the electric focusing module adjusts the lens angle to ensure that the projected image is horizontal (tilt ≤ 1°).

[0144] The system receives guidance information (text / graphic encoding) from the central platform, projects it onto a designated surface (wall or road surface), and sends the image of the projection effect back to the platform via a camera. If the platform determines that the image is "blurry", it will automatically recalibrate.

[0145] The two-way voice intercom module remains in standby mode. When the central platform initiates a call or on-site personnel trigger the microphone (such as by shouting for help), full-duplex communication is automatically enabled to achieve real-time dialogue between the platform and the site.

[0146] (iv) Environmental perception and feedback stage:

[0147] High-definition cameras and infrared thermal imagers continuously collect on-site data, and the edge computing module analyzes the data every 2 seconds to identify the status of fire, smoke, and personnel, and generate structured data (such as "Time 15:30, Location K2+300, Number of Personnel 2, No Fire").

[0148] The control unit summarizes the sensor analysis results and the working status of each module (such as the projector lamp temperature of 60℃), packages them, and sends them back to the central platform via the communication bus at a frequency of 1 time / second, supporting dynamic adjustment commands from the platform.

[0149] (v) Task completion and reset phase:

[0150] Upon receiving the "task completed" command from the central platform, the control unit shuts down the projection device and voice module, starts the drive motor, and the robot returns to the standby position (or designated recycling point) along the slide rail.

[0151] Upon arrival, the suspension movement mechanism locks, and the control unit enters a low-power standby mode, retaining only periodic self-test and communication functions, awaiting the next mission command.

[0152] In this embodiment, the system links with the tunnel's existing automatic fire alarm system and traffic incident detection system to receive alarm signals from the automatic fire alarm system and traffic incident detection system as the trigger source for starting the guidance process.

[0153] The positioning and communication infrastructure inside the tunnel uses an absolute position encoder based on a rigid slide rail to accurately position the tunnel-guided robot.

[0154] Furthermore, the tunnel positioning and communication infrastructure serves as the "data transmission and position reference center" for the rapid tunnel accident guidance system based on the visual guidance robot. It constructs an integrated support system centered on rigid slide rails, providing the tunnel guidance robot with millimeter-level precise positioning and millisecond-level low-latency communication. Simultaneously, it is compatible with the system's linkage requirements with existing tunnel equipment, making it a key support module ensuring efficient collaboration between the central control platform and the robot, and precise execution of commands. The following will elaborate on this infrastructure from an overall perspective to specific details:

[0155] I. Overall Function Overview:

[0156] The tunnel positioning and communication infrastructure undertakes two core tasks: First, precise positioning support, relying on an absolute position encoder based on rigid slide rails to output the absolute coordinates of the tunnel-guided robot in real time (error ≤ 0.5 meters), providing a position benchmark for the central control platform to schedule the robot and for the robot to accurately reach the accident point; Second, reliable communication assurance, using the rigid slide rails as a communication bus to build a low-latency communication network covering the entire tunnel (communication latency ≤ 0.5 seconds), enabling two-way data interaction between the central control platform and the robot (including command issuance, status feedback, and video stream transmission); At the same time, this infrastructure must be adaptable to the complex tunnel environment (such as electromagnetic interference, humidity changes, and dust) to ensure continuous positioning data and stable communication links, providing a fundamental guarantee for the overall system operation of "accurate positioning and no data loss";

[0157] II. Core Components and Functions:

[0158] (a) Absolute position coding positioning subsystem:

[0159] As the core carrier of the positioning function, it relies entirely on rigid slide rails to achieve the integration of physical scale and data reading, specifically including:

[0160] Rigid rail integrated absolute position encoder:

[0161] Structural design: Embedded installation along the inner wall of the rigid slide rail, adopting a "metal grating scale + photoelectric reading" structure, with a grating scale resolution of 0.1 mm / division, and the scale of the slide rail is continuous without any breaks throughout the tunnel (marked with linear coordinates from the tunnel entrance to the exit, such as "K0+000.000" to "K5+000.000"), ensuring that the absolute position can be read throughout the robot's movement without cumulative error;

[0162] Data Output: The encoder has built-in photoelectric sensors and signal processing chips. When the suspension movement mechanism of the tunnel guide robot moves along the slide rail, the photoelectric sensors capture the changes in the grating scale in real time, converting the physical position into digital coordinate signals (such as "K2+300.500", which means 2 kilometers, 300 meters and 50 centimeters in the tunnel). The signals are output to the robot's positioning assistance module at a frequency of 10Hz through the RS485 interface, and simultaneously uploaded to the position calibration unit of the central control platform.

[0163] Environmental adaptability: The encoder housing is made of 304 stainless steel with an IP65 protection rating. It can withstand a temperature range of -20℃ to 60℃, 95% humidity and dust in the tunnel, avoiding scale wear or signal distortion caused by environmental factors.

[0164] Positioning signal processing unit:

[0165] Signal calibration: Receive the raw coordinate signal output by the encoder, and correct signal deviations (such as scale offset caused by thermal expansion and contraction of the slide rail) through the "multi-node comparison" algorithm (selecting reference scale points at fixed locations in the tunnel, such as one reference point every 500 meters) to ensure that the coordinate data is completely matched with the actual physical location of the tunnel.

[0166] Redundancy backup: Built-in dual signal acquisition channels. When the main channel cannot read data due to sensor failure (such as phototube damage), it automatically switches to the backup channel (switching time ≤ 0.1 seconds) and sends a "positioning channel switching alarm" to the central control platform to ensure uninterrupted positioning data.

[0167] (ii) Rigid slide rail communication bus subsystem:

[0168] Using a rigid slide rail as the physical carrier, the communication function is integrated with the slide rail structure, avoiding additional wiring. Specifically, this includes:

[0169] Slide rail integrated communication conductor:

[0170] Structural design: Two sets of copper communication conductors (used for transmitting and receiving signals respectively) are embedded inside the rigid slide rail. The conductor surfaces are nickel-plated for rust prevention, and a polytetrafluoroethylene insulation layer is used to isolate them from the slide rail body to ensure that the communication signal and the slide rail power supply signal (the "power supply bus" as described in claim 3) do not interfere with each other. The communication conductors are laid continuously along the entire length of the slide rail without any joints or breaks to avoid signal attenuation.

[0171] Communication Protocol and Speed: Adopts industrial-grade CANopen protocol (compatible with anti-interference requirements of tunnel industrial environment), communication speed reaches 1Mbps, supports parallel transmission of "command frame + data frame + video frame" - among which command frame (such as movement command issued by central platform) has the highest priority and transmission delay ≤0.3 seconds; video frame (such as the on-site image returned by robot) is transmitted after H.265 compression, bandwidth usage ≤2Mbps, ensuring no lag when multiple robots transmit data at the same time;

[0172] Communication relay and amplification module:

[0173] Deployment logic: For tunnels longer than 1 kilometer, a relay amplifier module (IP65 protection level) is installed on the side of the slide rail every 500 meters. The module is connected to the communication conductor of the slide rail through brushes. After receiving the attenuated signal from upstream, it is amplified (amplification gain ≥20dB) and noise filtered (using a low-pass filter to filter interference signals above 100MHz) before being transmitted downstream to ensure stable communication signal strength throughout the tunnel (receiver signal strength ≥-80dBm).

[0174] Self-healing: The module has a built-in link detection function that checks the communication status of upstream and downstream every second. If a section of the slide rail communication conductor fails (such as breaking), the module automatically disconnects the faulty section and establishes temporary communication with the adjacent module through a backup wireless channel (LoRa protocol, transmission distance of 1 kilometer) to ensure that the communication link is not interrupted. At the same time, it sends "communication link fault location" information (such as "K2+000-K2+500 section slide rail communication conductor failure") to the central platform.

[0175] Two-way data interaction interface:

[0176] Interface with the central control platform: It uses the TCP / IP protocol to connect to the central control platform through an industrial switch, supporting bidirectional interaction of "command issuance" (such as emergency guidance commands and scheduling commands) and "data feedback" (such as robot position data and environmental perception data). The interface is redundantly designed (dual network port backup) to avoid communication interruption caused by single interface failure.

[0177] Interfacing with the robot: The robot's communication bus is physically connected by a flexible brush (made of wear-resistant graphite with a service life of ≥100,000 slides) on the robot's suspended moving mechanism, which contacts the communication conductor on the slide rail. The brush has a built-in anti-spark design to avoid arcing due to poor contact, and is adapted to the dynamic contact requirements during the robot's movement.

[0178] (III) Condition monitoring and fault self-healing subsystem:

[0179] Ensuring continuous and reliable positioning and communication functions includes:

[0180] Infrastructure Status Monitoring Unit:

[0181] Data Acquisition: Real-time acquisition of three types of core data: first, positioning subsystem data (encoder signal strength, reference point comparison deviation); second, communication subsystem data (communication bus signal attenuation rate, relay module operating temperature, link error rate); and third, environmental data (tunnel temperature, humidity, dust concentration), with an acquisition frequency of 1 time / second.

[0182] Anomaly warning: Set threshold judgment logic (such as communication error rate > 1%, encoder deviation > 0.3 mm). When the monitored data exceeds the threshold, automatically generate "anomaly warning information" (including anomaly type, location, and severity) and push it to the operation and maintenance interface of the central control platform. At the same time, store the anomaly log locally (storage period ≥ 1 year).

[0183] Fault self-healing execution module:

[0184] Location fault handling: If a section of the encoder fails (such as grating scale wear), the "adjacent encoder interpolation calculation" will be automatically activated (based on the coordinate data of the normal encoders before and after the faulty section, the position of the robot in the faulty section will be estimated by linear interpolation with an error of ≤1 meter) until the fault is repaired;

[0185] Communication failure handling: In addition to the temporary wireless channel switching of the relay module, if the entire sliding rail communication bus fails, the backup 4G / 5G industrial router (with built-in tunnel-specific high-gain antenna and received signal strength ≥-90dBm) will be automatically activated to switch the communication link to wireless mode, ensuring the transmission of core commands (such as robot emergency evacuation commands), with a switching time of ≤1 second.

[0186] III. Key Technology Principles:

[0187] (I) Absolute Position Positioning Principle: Physical scale + real-time reading, no cumulative error:

[0188] Unlike traditional incremental encoders (which rely on counting and accumulation, and are prone to cumulative errors due to slippage), this system uses an absolute position encoder based on a rigid slide rail. By engraving unique physical grating scales on the slide rail (each 0.1 mm corresponds to a unique encoding value), the photoelectric sensor directly reads the encoding value of the current position when the robot moves, eliminating the need for cumulative calculation and fundamentally eliminating cumulative errors. At the same time, through coordinate calibration of fixed reference points (such as evacuation exits and fire hydrant locations in tunnels), it ensures that the encoding value perfectly matches the actual geographical coordinates of the tunnel (such as "2300 meters from the entrance"), achieving a positioning accuracy of ±0.5 meters, meeting the needs of precise robot docking and collaborative scheduling.

[0189] (II) Slide rail communication bus principle: integrated design + anti-interference optimization, low latency and high reliability:

[0190] Functional reuse and isolation: The rigid slide rail simultaneously serves as both a "power supply bus" (claim 3) and a "communication bus". Through physical isolation (insulation layer separating the power supply conductor and the communication conductor) and signal isolation (power supply uses DC 24V, and communication uses differential signal), electromagnetic interference generated by the power supply current is prevented from affecting the communication signal. At the same time, it eliminates the need to lay communication cables separately in traditional systems, reducing construction complexity and the number of fault points.

[0191] Anti-interference design: The communication signal adopts differential transmission (CANopen protocol feature), which can cancel the common mode interference generated by motors, frequency converters and other equipment in the tunnel; the low-pass filter of the relay module further filters high-frequency electromagnetic noise (such as interference above 100MHz generated by vehicle spark plugs), ensuring that the communication bit error rate is ≤0.1%, meeting the requirement of "zero loss" of commands in emergency scenarios.

[0192] (III) Fault self-healing principle: Multi-level redundancy + fast switching ensures continuity:

[0193] The system employs a dual-protection approach combining hardware and algorithm redundancy: On the hardware side, the positioning subsystem features dual signal channels, the communication subsystem has dual network ports / backup wireless channels, and the relay module is redundantly deployed to ensure that a single hardware failure does not affect the overall functionality; on the algorithm side, the interpolation calculation for positioning failures and the link switching for communication failures are automatically executed through preset logic without manual intervention, with a switching time of ≤0.1 seconds, ensuring that positioning and communication functions remain "uninterrupted even when interrupted," thus meeting the "uninterrupted" guidance requirements in tunnel accident scenarios.

[0194] IV. Infrastructure workflow:

[0195] (a) Initialization phase:

[0196] After the system is powered on, the status monitoring unit first performs a self-test on the absolute position encoder, communication relay module, and interactive interface: it checks whether the encoder photoelectric sensor is normal (output signal strength ≥ 5V), whether the communication bus is connected (signal attenuation rate ≤ 5%), and whether the relay module's operating voltage is stable (24V±1V). If the self-test passes, it sends an "infrastructure ready" signal to the central control platform; if the self-test fails (e.g., a certain encoder has no signal), it sends "fault location information" (e.g., "encoder fault at K1+500").

[0197] The positioning signal processing unit loads the tunnel reference point coordinate data (such as "K0+000=tunnel entrance, K1+000=1 evacuation exit") and completes the mapping calibration between the encoded value and the actual coordinate; the communication subsystem initializes the CANopen protocol parameters (such as baud rate 1Mbps, frame format) and establishes the communication link between the central platform and the robot.

[0198] (II) Location data acquisition and transmission stage:

[0199] When the tunnel-guided robot moves along the slide rail, the robot's positioning assistance module reads the grating scale of the absolute position encoder on the slide rail through the brush and transmits the original encoded value to the positioning signal processing unit.

[0200] The positioning signal processing unit calibrates the original encoded value (compared with the reference point to correct the deviation) and generates a positioning data packet of "robot ID + current coordinates + timestamp". This data packet is uploaded to the central control platform in real time via the communication bus (upload frequency 10Hz) and simultaneously fed back to the robot for position correction (such as adjusting the moving speed to ensure accurate docking).

[0201] (III) Communication Data Interaction Phase:

[0202] Command Issuance: Emergency guidance and scheduling commands generated by the central control platform are transmitted to the communication subsystem via the TCP / IP interface. The communication subsystem encapsulates the commands into CANopen command frames (marked with the target robot ID) and transmits them downwards via the slide rail communication bus. The relay module amplifies and filters the command frames to ensure that the commands accurately reach the target robot. After receiving the commands via the brush, the robot sends a "command reception confirmation" to the platform.

[0203] Data feedback: The robot's environmental perception data (such as on-site video, infrared thermal imaging) and status data (such as motor current, projection status) are compressed using H.265 and encapsulated into CANopen data frames, which are then transmitted back to the central platform via the communication bus. If the data volume is large (such as 4K video), the communication subsystem automatically adopts a "fragmented transmission + retransmission mechanism" to ensure data integrity (retransmission count ≤ 3 times, and if unsuccessful, switch to the backup wireless channel).

[0204] (iv) Abnormal handling phase:

[0205] The status monitoring unit monitors the data in real time. If it finds that "the encoder deviation of a certain segment is greater than 0.3 mm", it immediately initiates positioning fault handling: it activates the backup channel in the dual signal channel. If the backup channel is normal, it switches to the backup channel; if the backup channel is also faulty, it starts interpolation calculation, estimates the robot position by using the encoder data before and after the faulty segment, and sends a "positioning accuracy reduction warning" to the platform (such as "the positioning error of segment K2+000-K2+500 has increased to 1 meter").

[0206] If the communication bus signal attenuation rate is greater than 20% (e.g., the sliding rail communication conductor breaks), the relay module immediately disconnects the faulty section, starts the backup LoRa wireless channel, establishes a temporary communication link between adjacent modules, and sends a "communication link switching alarm" to the platform; if the entire bus fails, the backup 4G / 5G router is activated to ensure that the core command transmission is not interrupted.

[0207] (V) Conclusion:

[0208] Once the tunnel accident is resolved, the central control platform issues a "system standby" command. The infrastructure stops high-frequency acquisition of positioning data (reduced to once per minute), the communication subsystem switches to low-power mode (reducing the power of the relay module), and the status monitoring unit only retains monitoring of key parameters (such as voltage and temperature).

[0209] The status monitoring unit generates an "infrastructure operation report," which includes the positioning accuracy, communication latency, number of failures, and handling results during this incident. The report is stored in a local database and uploaded to the central platform to provide a basis for subsequent operation and maintenance optimization.

[0210] The present invention discloses a rapid guidance system for tunnel accidents based on a visual guidance robot, the operation steps of which are as follows:

[0211] I. Process Initiation: Multi-source Alarm Information Reception and Verification (Triggering Phase):

[0212] The core of this phase is to accurately identify valid incidents and avoid false triggers through the linkage between the system and existing equipment. This includes the linkage design between the corresponding system and the tunnel's existing automatic fire alarm system and traffic incident detection system, as well as the "multi-source information receiving unit" function of the central control platform.

[0213] Alarm signal acquisition: Existing systems within the tunnel (automatic fire alarm system, traffic incident detection system) and the tunnel monitoring system send signals to the central control platform in real time.

[0214] Automatic fire alarm system: When the smoke concentration in the tunnel is ≥0.1mg / m³ and the temperature is ≥60℃, a "fire alarm + specific location (e.g., K2+300)" signal is sent.

[0215] Traffic incident detection system: Through video analysis or loop detectors, identify vehicle collisions and abnormal parking (delay ≥30 seconds) and send a "traffic accident + lane information (e.g., left lane)" signal;

[0216] Tunnel monitoring system: When cameras capture images of people lingering or vehicles congesting, they simultaneously upload the video stream and associated coordinates to the platform;

[0217] Signal cross-validation: The central control platform performs "same-source comparison + logical judgment" on multi-source signals to eliminate false alarms.

[0218] If the automatic fire alarm system sends a "K2+300 fire" signal, the platform will simultaneously retrieve the surveillance video of that location (to determine if there is smoke) and traffic incident detection data (to determine if there are any vehicles stopped). If both types of data confirm a fire, it will be considered a "valid alarm".

[0219] If only a single system sends a signal (e.g., only the monitoring system detects suspected smoke, but the fire alarm system does not respond), the platform marks it as "pending verification" and pushes it to the management personnel interface. After manual review, it is confirmed whether to start the guidance process.

[0220] Guidance Level Determination: The central control platform matches a preset guidance level based on the valid alarm type.

[0221] Level 1 Emergency Guidance (Fire, Explosion): Triggered with the highest priority, requiring robot dispatch to be initiated within 10 seconds;

[0222] Level 2 Regular Guidance (Vehicle Collision, Minor Congestion): Triggers regular priority, initiates dispatch within 20 seconds to ensure reasonable allocation of resources;

[0223] II. Instruction Generation and Robot Scheduling (Central Decision-Making Stage):

[0224] This stage is the core function of the central control platform. It needs to generate guidance instructions based on accident information and select robots through intelligent algorithms. This corresponds to the core function of the central control platform in generating emergency instructions, the algorithm design of the intelligent scheduling module, and the support role of the tunnel's positioning and communication infrastructure in providing accurate location data.

[0225] Emergency guidance instruction generation: The central control platform loads static tunnel parameters (length, evacuation exit locations, number of lanes) and real-time accident data (type, location, impact range) to generate dual-mode guidance instructions (text + graphics).

[0226] Text instructions, such as "Fire ahead at K2+300, evacuate to K2+200 evacuation exit by following the green arrow on the right," are quickly recognized by the appropriate personnel.

[0227] Graphical instructions: such as a 20cm wide green evacuation arrow (pointing to the exit) and a red no-entry sign (marking the accident area), with a resolution adapted to the display requirements of the robot's high-lumen laser projector; the instructions are encoded in a format that the robot can parse (text UTF-8 encoding, graphic pixel matrix data), waiting to be issued synchronously with the scheduling instructions;

[0228] Robot Status Screening and Optimal Deployment: The central control platform obtains the real-time status of all tunnel-guided robots through the tunnel's positioning and communication infrastructure.

[0229] Basic status: location (e.g., robot A is at K1+000, robot B is at K3+000), battery level (≥50% is available), task status (idle / in execution);

[0230] Functional status: Whether the projection device, two-way voice module, and environmental perception sensors are functioning properly (judged by the "health report" returned by the robot control unit); the intelligent scheduling module calculates the optimal solution based on the principle of "shortest travel time + complete functionality":

[0231] Single robot scheduling: If the accident impact range is ≤50 meters (such as a minor traffic accident), select the robot that is "closest to the accident point (e.g., robot A is only 200 meters from K2+300, and the travel time = 200 meters / 0.5 m / s = 400 seconds) and has a battery level ≥50%";

[0232] Multi-robot collaboration: If the impact range of an accident is ≥100 meters (such as a fire), dispatch 2 or more robots (e.g., robot A is responsible for guiding from K2+200 to K2+300, and robot B is responsible for guiding from K2+300 to K2+400), and allocate the slide rail section through "time window scheduling" (e.g., A moves to K2+300 in 0-400 seconds, and B moves to K2+400 in 200-600 seconds) to avoid collisions;

[0233] Command issuance and confirmation: The central control platform issues "scheduling instructions (target location, task assignment) + guidance instructions (text / graphics)" to the selected robot through the "rigid sliding rail communication bus" of the positioning and communication infrastructure in the tunnel. After receiving the instructions, the robot sends back a "command reception confirmation" signal through the communication bus. If no confirmation is received within 10 seconds, the platform reissues the instructions (up to 3 times) to ensure that the instructions are delivered.

[0234] III. Robot Movement and Localization (Execution Preparation Phase):

[0235] This stage involves the robot moving from its standby position to the accident site. Stability and accuracy are ensured primarily by rigid guide rails and positioning infrastructure. This corresponds to the structural design of the tunnel-guided robot's suspension movement mechanism, the positioning function of the absolute position encoder, and the design characteristics of the robot's modular functional cabin.

[0236] Robot Start-up and Movement: After parsing the scheduling instructions, the robot control unit starts the drive motor, which drives the power wheel set of the suspension movement mechanism to move along the rigid slide rail.

[0237] The drive wheel set uses a "gear meshing" structure to clamp the slide rail, and the driven wheel set is pre-tightly pressed against the side of the slide rail by a spring to ensure smooth movement (vibration amplitude ≤ ±2°).

[0238] The drive motor has stepless speed regulation (e.g., moving at a speed of 0.8 m / s), the encoder provides real-time feedback on the moving distance, and the positioning auxiliary module synchronously reads the absolute position encoder information on the rigid slide rail (resolution 0.1 mm / division), and reports the current coordinates (e.g., "K2+100.500") to the central platform every 0.5 seconds.

[0239] Safety safeguards activated: The anti-fall locking mechanism of the suspended moving mechanism is monitored throughout the process.

[0240] If a sudden drop of 50% in suspension tension is detected (e.g., abnormal force caused by local deformation of the slide rail) or the moving speed exceeds 1.5 times the rated speed (e.g., motor malfunction), the metal claws will automatically lock within 0.1 seconds to fix the robot to the slide rail.

[0241] At the same time, the robot sends an "abnormal alarm + current location" to the central platform, which immediately suspends scheduling and restarts after the fault is resolved.

[0242] Precise docking: When the robot approaches the target position (e.g., 10 meters away from K2+300), the positioning assistance module sends a deceleration signal to the control unit, and the drive motor slows down to 0.2 m / s; after reaching the target position, the driven wheel group locks (to prevent slippage), and the robot sends a "in position" signal to the platform, ready to perform the guidance task;

[0243] IV. On-site guidance and environmental awareness (core execution phase):

[0244] This stage is the core of the robot's "visual guidance + voice interaction + environmental monitoring" process. It corresponds to the core configuration of the tunnel-guided robot: a visual projection device, a two-way voice intercom module, and environmental perception sensors. It also utilizes the display characteristics of a high-lumen laser projector and the collaborative perception capabilities of a high-definition camera, an infrared thermal imager, and an edge computing module.

[0245] Visualized guided execution: The robot control unit receives guidance instructions from the central platform and activates the visualization projection device.

[0246] Optical path calibration: The built-in gyroscope detects the body posture, and the electric focusing module adjusts the lens angle (±15° range) to ensure that the projected image is horizontal (tilt ≤1°); if the projection surface is a tunnel wall (concrete material), the contrast is automatically adjusted to 80%; if it is a road surface (asphalt material), it is adjusted to 100% to ensure that the arrow is clearly visible (visible from 50 meters away).

[0247] Content projection: Project the encoded text / graphics onto the designated surface, such as projecting a "green evacuation arrow" onto the right wall of K2+300, and projecting a red "No Entry" sign onto the road surface. The flashing frequency of the arrows is set according to the guidance level (2Hz for level 1 guidance and 1Hz for level 2 guidance).

[0248] Two-way voice interaction activated: The two-way voice intercom module enters standby mode.

[0249] If the central platform needs to make announcements to personnel on site (such as "Please stay away from the accident vehicle and evacuate according to the arrows"), the loudspeaker (output power 10W, coverage radius 15 meters) will emit sound in a directional manner to filter airflow noise in the tunnel (noise resistance ≥80dB).

[0250] If on-site personnel call for help (such as "someone is trapped"), the high-sensitivity microphone (with a pickup distance of 3-10 meters) will capture the sound and automatically start full-duplex communication with a voice delay of ≤0.3 seconds, enabling real-time dialogue between the platform and the site.

[0251] Environmental sensing and data feedback: Environmental sensing sensors continuously collect on-site data.

[0252] High-definition cameras (20 megapixels, 120° field of view) capture accident footage, while infrared thermal imagers (640×512 resolution) can penetrate smoke to identify personnel silhouettes (even in visibility <5 meters).

[0253] The edge computing module runs the YOLOv5s model, analyzes the video stream within 1 second, automatically identifies fires (accuracy ≥ 95%) and people stranded (accuracy ≥ 98%), and generates structured results (such as "2 stranded people at K2+300, no high temperature points").

[0254] After being compressed using H.265 (compression ratio 10:1), the data is transmitted back to the central platform via the rigid slide rail communication bus at a frequency of 1 time / second, providing a basis for the platform's dynamic adjustment commands.

[0255] V. Fault Handling and Dynamic Adjustment (Collaborative Optimization Phase):

[0256] This phase addresses unforeseen malfunctions during operation (such as robot failures or communication interruptions) by employing system redundancy and self-healing capabilities to ensure uninterrupted workflow. This includes the functional design of rigid guide rails serving as both power supply and communication buses, strategies for multi-robot collaborative scheduling, and the ease of maintenance through modular robot design.

[0257] Robot Fault Handling: If a robot (e.g., robot A) suddenly malfunctions (e.g., the projection device fails), the central platform identifies the problem through status feedback data:

[0258] Redundancy switching: The intelligent scheduling module immediately selects a backup robot (such as robot C, located at K1+500), issues a scheduling instruction, and robot C moves along the slide rail to the task area of ​​robot A to take over the guiding task;

[0259] Fault isolation: The faulty robot automatically activates the "safety lock", the fall protection mechanism stabilizes the body, and at the same time sends the "fault type + location" to the platform to facilitate subsequent maintenance;

[0260] Communication interruption handling: If there is a partial fault in the rigid slide rail communication bus (such as a conductor breakage in the K2+200-K2+300 segment):

[0261] The relay module automatically disconnects the faulty segment and establishes temporary communication through the LoRa wireless channel (transmission distance 1 km); if the entire bus is interrupted, the robot starts the "local storage-forward mode" to temporarily store the environmental perception data in the local cache (capacity ≥10GB), and forwards it to the platform in chronological order after communication is restored to ensure that the data is not lost.

[0262] The robot control unit initiates a preset emergency procedure, projecting fixed guidance commands (such as "evacuate to the tunnel entrance") until communication is restored;

[0263] Dynamic command adjustment: The central platform optimizes the guidance strategy based on the environmental data transmitted back by the robot.

[0264] If the infrared thermal imager detects that the fire is spreading towards K2+400, the platform issues an instruction to robot B to "adjust the projection position to K2+350, with the arrow pointing to the evacuation exit at K2+300".

[0265] If the number of people on site increases (e.g., 5 stranded people are detected), the platform will dispatch additional robots (e.g., robot D) to provide support and increase the frequency of voice broadcasts to ensure comprehensive guidance coverage.

[0266] VI. Task Completion and System Reset (Final Stage):

[0267] This phase, following the completion of accident handling, involves system recovery and data archiving, corresponding to the data recording unit function of the central control platform and the robot's standby logic:

[0268] End command trigger: When tunnel management personnel confirm that the accident has been handled (such as fire extinguishing and vehicle towing away), they issue a "task end" command through the central platform, or the robot edge computing module does not detect personnel / abnormalities for 5 consecutive minutes and automatically sends a "no abnormality on site" signal to the platform to trigger the end process;

[0269] Robot reset: After the robot receives the end command:

[0270] The projection device and voice module are turned off, the drive motor starts, and it returns to the preset standby position (such as tunnel entrance K0+000) along the slide rail. During the movement, the positioning assistance module continuously provides feedback on the position to ensure accurate stopping.

[0271] Upon reaching the standby position, the suspension movement mechanism locks, and the control unit enters a low-power mode, retaining only periodic self-checks (once every 10 seconds) and communication functions, waiting for the next task;

[0272] Data archiving and report generation: The central control platform aggregates data from the entire process.

[0273] Record content: alarm time / type, robot scheduling record (number, movement path, task assignment), fault handling process, key environmental perception data;

[0274] Report generation: Automatically generate a "Tunnel Accident Guidance Report", which includes indicators such as guidance response time (e.g., 45 seconds from alarm to robot deployment), coverage area, and personnel evacuation efficiency. The report is pushed to the administrator's email address and the data storage period is ≥1 year, providing a basis for system optimization.

[0275] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A rapid guidance system for tunnel accidents based on a visual guidance robot, characterized in that: It includes a central control platform, a rigid slide rail fixedly installed at the top of the tunnel and extending through the entire length of the tunnel, at least one tunnel guiding robot that can move along the rigid slide rail, and positioning and communication infrastructure within the tunnel. The central control platform is used to receive accident alarm information sent by the tunnel monitoring system, generate emergency guidance instructions, and dispatch tunnel guidance robots. The tunnel guiding robot includes a drive motor, a power supply system, a wireless communication module, a visualization projection device, a two-way voice intercom module, an environmental perception sensor, and a robot control unit. The drive motor is used to drive the suspension movement mechanism, so that the tunnel guiding robot moves along a rigid slide rail. The visualization projection device is used to project emergency guidance instructions issued by the central control platform onto the tunnel wall or road surface in the form of text or graphics. The two-way voice intercom module is used to establish a full-duplex voice communication link between the central control platform and personnel at the accident site. The robot control unit is electrically connected to the drive motor, power supply system, wireless communication module, visualization projection device, two-way voice intercom module and environmental perception sensor, respectively, and is used to receive and execute instructions from the central control platform. The tunnel positioning and communication infrastructure provides the tunnel-guided robot with accurate indoor positioning information and reliable low-latency communication.

2. The tunnel accident rapid guidance system based on a visual guidance robot according to claim 1, characterized in that: The suspension movement mechanism includes a power wheel set, a driven wheel set, and a fall arrest mechanism. The power wheel set is driven by the drive motor and is clamped or engaged on the rigid slide rail. The fall arrest mechanism automatically locks when it detects loss of force or overspeed, thus fixing the tunnel guide robot to the rigid slide rail.

3. The tunnel accident rapid guidance system based on a visual guidance robot according to claim 1, characterized in that: The rigid slide rail serves as both a power supply bus and a communication bus, continuously supplying power to the tunnel-guided robot through its brushes or inductive power extraction module, and interacting with the central control platform via bus communication.

4. A rapid tunnel accident guidance system based on a visual guidance robot according to claim 1, characterized in that: The central control platform also includes an intelligent scheduling module, which is used to automatically calculate the optimal dispatch plan based on the accident location, the real-time location and status of the tunnel guiding robot, and can schedule multiple tunnel guiding robots to operate collaboratively on a single rigid slide rail.

5. A rapid tunnel accident guidance system based on a visual guidance robot according to claim 1, characterized in that: The environmental perception sensors include a high-definition camera and an infrared thermal imager. The robot control unit is also equipped with an edge computing module for real-time analysis of the video stream, automatic identification of fire, smoke or personnel lingering status, and real-time transmission of analysis results.

6. A rapid tunnel accident guidance system based on a visual guidance robot according to claim 1, characterized in that: The system is linked with the tunnel's existing automatic fire alarm system and traffic incident detection system, receiving alarm signals from the automatic fire alarm system and the traffic incident detection system as the trigger source for initiating the guidance process.

7. A rapid tunnel accident guidance system based on a visual guidance robot according to claim 1, characterized in that: The positioning and communication infrastructure inside the tunnel uses an absolute position encoder based on the rigid slide rail to accurately position the tunnel guiding robot, and then performs secondary calibration with the positioning module mounted on the track to achieve precise positioning.

8. A rapid tunnel accident guidance system based on a visual guidance robot according to claim 1, characterized in that: The tunnel guiding robot adopts a modular functional cabin design. The visualization projection device and the two-way voice intercom module can be controlled by electric push rods or servo motors to perform lifting or rotating movements in order to find the best projection and sound reception angles.

Images