Intelligent rubber-tyred vehicle in tunnel

Abstract

The invention relates to the technical field of tunnel intelligent construction equipment, and discloses an in-tunnel intelligent rubber-tyred vehicle which comprises a vehicle frame, the two ends of the vehicle frame are fixedly connected with two cabs, and the upper surfaces of the cabs are fixedly connected with a gateway switch and a vehicle-mounted industrial personal computer; a multi-mode sensor module is arranged outside the cab, and the multi-mode sensor module comprises a laser radar, a visual camera and an ultrasonic probe and is used for sensing the environment in the tunnel; the gateway switch is internally provided with a fusion positioning module and a wireless communication module and is used for realizing vehicle positioning and remote communication; a control module and an execution module are arranged in the vehicle-mounted industrial personal computer, and the vehicle-mounted industrial personal computer is used for receiving the sensing information, generating a control instruction and then driving the vehicle to run according to the control instruction. The heterogeneous data of the laser radar, the visual camera and the ultrasonic probe are fused through the multi-mode sensing system, a tunnel three-dimensional dynamic environment model is constructed, and the obstacle type and spatial distribution are recognized in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent tunnel construction equipment, and in particular to an intelligent rubber-wheeled vehicle in a tunnel. Background Art

[0002] In traditional tunnel construction, rubber-tyred vehicles mostly rely on manual driving or simple remote control operation, which has significant limitations in narrow, closed tunnel environments with high dust concentrations and frequent dynamic obstacles.

[0003] In existing technologies, vehicle perception systems often use a single lidar or camera solution, which has difficulty coping with sudden changes in lighting in tunnels, multipath reflections from metal structures, and interference from construction dust, resulting in an increased rate of missed obstacle detection; communication modules are mostly based on single-band transmission, and are prone to control command delays or interruptions due to signal attenuation from concrete walls and electromagnetic interference from large machinery, restricting the reliability of remote emergency intervention; energy systems generally use lead-acid battery packs, which have problems such as low energy density, poor temperature adaptability, and short cycle life, and cannot meet the all-weather power supply needs under continuous tunnel excavation conditions.

[0004] In addition, traditional equipment lacks multi-sensor data fusion and edge computing capabilities, and dynamic path planning is not real-time enough, which can easily lead to safety risks when encountering falling rocks, temporary fences, or construction workers crossing. At the same time, the continuous operation of high-power consumption equipment exacerbates the risk of battery thermal runaway, making it difficult to adapt to the technical requirements of modern tunnel projects for intelligent and green construction equipment. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides an intelligent rubber-wheeled vehicle for use in tunnels, which solves the problems of environmental perception blind spots, communication interruption risks, dynamic obstacle avoidance delays, insufficient energy endurance and low human-machine collaboration efficiency existing in traditional tunnel rubber-wheeled vehicles in complex and closed environments.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent rubber-tyred vehicle in a tunnel, comprising a frame, two cabs fixedly connected to both ends of the frame, and a gateway switch and an on-board industrial computer fixedly connected to the upper surface of the cab;

[0007] A multimodal sensor module is provided on the outside of the cab. The multimodal sensor module includes a laser radar, a visual camera and an ultrasonic probe for sensing the environment inside the tunnel;

[0008] The gateway switch is internally provided with a fusion positioning module and a wireless communication module for realizing vehicle positioning and remote communication;

[0009] The vehicle-mounted industrial computer is internally provided with a control module and an execution module for receiving sensing information and generating control instructions, and then driving the vehicle according to the control instructions;

[0010] A remote control terminal is provided inside the cab for monitoring vehicle status and realizing remote operation.

[0011] Preferably, the laser radar includes four 32-line laser radars and two blind spot laser radars. The four 32-line laser radars are respectively installed at the rearview mirrors of the two cabs, and the two blind spot laser radars are installed in the middle of the front end of the two cabs.

[0012] Preferably, the visual camera includes two binocular cameras and fourteen 360 fisheye cameras, the two binocular cameras are installed in the middle of the front end of the two cabs, the binocular camera is located directly above the blind spot laser radar, and the fourteen 360 fisheye cameras are respectively installed on the upper side of the front end of the two cabs, the four corners of the front end of the two cabs and the upper end of the side surfaces of the two cabs.

[0013] Preferably, the ultrasonic probes are installed on the bumpers of the two cabs and twelve are installed in total. The detection distance of the ultrasonic probes is 0.1-3m and the accuracy is ±2cm.

[0014] Preferably, the fusion positioning module includes a UWB positioning unit and an IMU inertial navigation unit, and the fusion positioning module achieves high-precision positioning through data fusion.

[0015] Preferably, the wireless communication module supports dual-frequency transmission in the 2.4 GHz and 5.8 GHz frequency bands, and the communication delay is less than 50 milliseconds.

[0016] Preferably, the control system is built based on an edge computing architecture, and the on-board industrial computer is used to fuse multi-sensor data and generate control decisions.

[0017] Preferably, the control module includes a path planning algorithm unit and an obstacle avoidance unit, and the obstacle avoidance module realizes dynamic path adjustment based on deep learning target recognition.

[0018] Preferably, a lithium iron phosphate battery is installed in the vehicle frame, and a battery management unit is provided in the control module for monitoring the battery status.

[0019] Preferably, the remote control terminal includes a video monitoring unit, an operation instruction unit and a parameter trajectory echo unit, which are used to realize human-computer interaction and remote control functions.

[0020] The present invention provides an intelligent rubber-wheeled vehicle for use in tunnels. It has the following beneficial effects:

[0021] 1. The present invention uses a multimodal perception system to fuse heterogeneous data from lidar, visual cameras, and ultrasonic probes to construct a three-dimensional dynamic environment model of the tunnel and identify obstacle types and spatial distribution in real time. Combined with an on-board industrial computer with an edge computing architecture, it achieves a millisecond-level decision-making closed loop. In tunnel scenarios without GPS signals, dust interference, and dynamic changes in construction machinery, the vehicle's autonomous navigation path planning accuracy and sudden obstacle avoidance response speed are improved, significantly enhancing driving safety under complex working conditions.

[0022] 2. The present invention utilizes the dual-band redundant transmission mechanism of the wireless communication module and the human-computer interaction system of the remote control terminal to support low-latency video monitoring unit feedback and precise issuance of operation instruction units. When the autonomous driving mode is abnormal, the parameter trajectory echo unit can be used to quickly locate the fault node. Combined with manual remote intervention, seamless switching of vehicle control rights can be achieved, effectively ensuring the reliability of emergency control in extreme scenarios such as sudden power outages and communication interruptions in tunnels.

[0023] 3. The present invention optimizes the charge and discharge strategy and monitors the health status of the battery cells in real time through the coordinated control of the lithium iron phosphate battery and the battery management unit. It combines a multi-level energy consumption control mechanism to provide a stable power supply for high-load sensing equipment and drive mechanisms, and maintains 8 hours of continuous operation in a tunnel temperature environment ranging from -20°C to 60°C. At the same time, the filter circulation water purification system reduces the equipment's heat dissipation requirements, achieving a balance between efficient use of clean energy and all-weather endurance performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a side view of the present invention;

[0025] Figure 2 Schematic diagram of the cab of the present invention;

[0026] Figure 3 A side view of the cab of the present invention;

[0027] Figure 4 A top view of the cab of the present invention;

[0028] Figure 5 A schematic diagram of the sensor layout in the present invention;

[0029] Figure 6 is an illustration of the sensor layout of the present invention;

[0030] Figure 7 Schematic diagram of the control system of the present invention;

[0031] Figure 8 is a schematic diagram of a communication system in the present invention;

[0032] Figure 9 Flowchart of data processing in the present invention.

[0033] Among them, 1. Frame; 2. Cab; 3. Gateway switch; 4. On-board industrial computer; 5. Multimodal sensor module; 501. LiDAR; 502. Visual camera; 503. Ultrasonic probe; 301. Fusion positioning module; 302. Wireless communication module; 401. Control module; 402. Execution module; 6. Remote control terminal; 5011. 32-line LiDAR; 5012. Blind-filling LiDAR; 5021. Binocular camera; 5022. 360 fisheye camera; 3011. UWB positioning unit; 3012. IMU inertial navigation unit; 4011. Path planning algorithm unit; 4012. Obstacle avoidance unit; 7. Lithium iron phosphate battery; 4013. Battery management unit; 601. Video monitoring unit; 602. Operation instruction unit; 603. Parameter trajectory echo unit. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] Please see the attached Figure 1 -Attached Figure 9 The embodiment of the present invention provides an intelligent rubber-wheeled vehicle for use in tunnels, comprising a vehicle frame 1, two cabs 2 being fixedly connected to both ends of the vehicle frame 1, and a gateway switch 3 and an on-board industrial computer 4 being fixedly connected to the upper surface of the cabs 2;

[0036] A multimodal sensor module 5 is provided outside the cab 2. The multimodal sensor module 5 includes a laser radar 501, a visual camera 502, and an ultrasonic probe 503, which are used to sense the environment inside the tunnel.

[0037] The gateway switch 3 is internally provided with a fusion positioning module 301 and a wireless communication module 302 for realizing vehicle positioning and remote communication;

[0038] The vehicle-mounted industrial computer 4 is internally provided with a control module 401 and an execution module 402, which are used to receive sensing information and generate control instructions, and then drive the vehicle according to the control instructions;

[0039] A remote control terminal 6 is provided inside the cab 2 for monitoring the vehicle status and realizing remote operation.

[0040] Specifically, the frame 1 serves as an integral load-bearing structure to connect the cabs 2 at both ends and supports the integrated installation of other functional modules; a gateway switch 3 is fixed inside the cab 2, which transmits the data collected by the multimodal sensor module 5 to the on-board industrial computer 4 in real time to ensure seamless connection between perception and control instructions; the on-board industrial computer 4 serves as the core computing unit, which integrates the perception data of the laser radar 501, the visual camera 502 and the ultrasonic probe 503 based on the edge computing architecture, realizes path planning and obstacle avoidance decisions through the control module 401, and drives the execution module 402 to control the vehicle steering, braking and power system; in the multimodal sensor module 5, the laser radar 501 constructs a three-dimensional environment model of the tunnel through high-precision point cloud data, and the visual camera 502 captures image information based on binocular vision and fisheye lens to identify obstacle types and movement trends The ultrasonic probe 503 compensates for the blind spots of laser and vision through short-range detection, and the three work together to achieve centimeter-level environmental perception; the gateway switch 3 has an integrated fusion positioning module 301, which compensates for single positioning errors through data fusion of the UWB positioning unit 3011 and the IMU inertial navigation unit 3012, ensuring high-precision autonomous positioning in the tunnel without GPS signals; the wireless communication module 302 uses dual-frequency transmission technology to ensure low-latency interaction of video monitoring, operation instructions and trajectory echo data of the remote control terminal 6, supporting manual remote intervention; the remote control terminal 6 realizes visual monitoring and real-time control of the vehicle status by integrating the video monitoring unit 601, the operation instruction unit 602 and the parameter trajectory echo unit 603, and finally forms a closed-loop operation process of environmental perception-positioning decision-making-control execution-remote collaboration.

[0041] The laser radar 501 includes four 32-line laser radars 5011 and two blind spot laser radars 5012. The four 32-line laser radars 5011 are respectively installed at the rearview mirrors of the two cabs 2, and the two blind spot laser radars 5012 are installed at the middle of the front end of the two cabs 2.

[0042] Specifically, the laser radar 501 includes four 32-line laser radars 5011 and two blind-point laser radars 5012. The four 32-line laser radars 5011 generate high-density point cloud data through multi-beam scanning, construct a three-dimensional spatial model in the tunnel, and update the dynamic obstacle distribution in real time to support the accuracy requirements of global path planning; the two blind-point laser radars 5012 focus on the key area in front of the vehicle through high-frequency narrow field of view scanning, enhance the detection sensitivity to low obstacles or complex terrain, make up for the vertical field of view blind spot of conventional laser radar, and improve the recognition reliability of close-range obstacles; the 32-line laser radar 5011 and the blind-point laser radar 5012 optimize the continuity and redundancy of environmental perception in scenarios where the tunnel is curved or the lighting is insufficient through heterogeneous data fusion, provide stable and reliable spatial topology information for the control module 401, and ultimately realize omnidirectional and no-dead-angle perception of the vehicle in narrow and long enclosed spaces.

[0043] The visual camera 502 includes two binocular cameras 5021 and fourteen 360 fisheye cameras 5022. The two binocular cameras 5021 are installed in the middle of the front end of the two cabs 2. The binocular camera 5021 is located directly above the blind spot laser radar 5012. The fourteen 360 fisheye cameras 5022 are respectively installed on the upper side of the front end of the two cabs 2, the four corners of the front end of the two cabs 2 and the upper end of the side surfaces of the two cabs 2.

[0044] Specifically, the visual camera 502 includes two binocular cameras 5021 and fourteen 360 fisheye cameras 5022. The binocular camera 5021 obtains the precise distance information of the obstacle in front in real time through parallax calculation based on the principle of stereo vision, and combines the deep learning algorithm to identify the target type and motion state, providing semantic-level environmental understanding for dynamic obstacle avoidance; the 360 fisheye camera 5022 captures a panoramic image of the vehicle's surroundings through an ultra-wide-angle lens, uses distortion correction and multi-frame stitching technology to eliminate blind spots, and simultaneously monitors complex scenes such as side and rear obstacle intrusion, tunnel wall deformation or ground potholes, and assists in building a global environmental map; the binocular camera 5021 and the 360 fisheye camera 5022 use multi-source data fusion to complement the perception limitations caused by differences in lighting conditions, enhance the continuity of target tracking in scenes of alternating light and dark or dust interference in the tunnel, and ultimately provide the path planning algorithm unit 4011 with high-resolution visual features and semantic segmentation results, supporting the vehicle to achieve all-weather omnidirectional perception and safe navigation in narrow spaces.

[0045] The ultrasonic probes 503 are installed at the bumpers of the two cabs 2 and there are twelve ultrasonic probes installed in total. The detection distance of the ultrasonic probes is 0.1-3m and the accuracy is ±2cm.

[0046] Specifically, six ultrasonic probes 503 are installed on the bumpers of the two cabs 2. By emitting high-frequency sound waves and receiving reflected signals, they detect the outlines and relative distances of close-range obstacles within a range of 0.1-3 meters around the vehicle bumper in real time. Combined with the triangulation algorithm of the multi-probe array, the obstacle edge positioning accuracy is improved to ±2 cm, and small-sized targets such as gravel, low curbs, or temporary piles that are easily missed by the lidar 501 and visual camera 502 are accurately identified. The six probes use a time-staggered scanning strategy to avoid signal crosstalk. When dust or strong light in the tunnel causes the performance of the optical sensor to degrade, the six probes provide redundant obstacle distance information. At the same time, based on the short-range detection characteristics, emergency braking or low-speed creep control strategies are triggered. This forms a gradient synergy with the medium- and long-range perception of the lidar 501 and visual camera 502, ultimately enhancing the vehicle's close-range obstacle avoidance safety and real-time control response in reversing, narrow road passage, or complex construction scenarios, and improving the full-condition coverage capability of the multimodal perception system.

[0047] The fusion positioning module 301 includes a UWB positioning unit 3011 and an IMU inertial navigation unit 3012. The fusion positioning module 301 achieves high-precision positioning through data fusion.

[0048] Specifically, the fusion positioning module 301 includes a UWB positioning unit 3011 and an IMU inertial navigation unit 3012, wherein the UWB positioning unit 3011 receives the ultra-wideband signal of a preset base station in the tunnel, calculates the precise distance and angle information between the vehicle and the base station, and provides absolute position data in the global coordinate system; the IMU inertial navigation unit 3012 measures the vehicle's angular velocity and acceleration in real time through a three-axis accelerometer and gyroscope, and generates a short-term, high-precision relative displacement increment based on the dead reckoning algorithm; the two are synchronized in time and space and data fused through the Kalman filter algorithm, and the UWB absolute positioning is used to correct the IMU's cumulative error, while the IMU's high-frequency update characteristics are used to compensate for the positioning continuity when the UWB signal is blocked, ultimately achieving centimeter-level positioning accuracy in scenarios where there is no GPS signal in the tunnel and multipath interference is severe, providing a stable and reliable vehicle posture reference for path planning and dynamic obstacle avoidance, and supporting autonomous navigation capabilities under complex working conditions in the tunnel.

[0049] The wireless communication module 302 supports dual-band transmission in the 2.4 GHz and 5.8 GHz frequency bands, and the communication delay is less than 50 milliseconds.

[0050] Specifically, the wireless communication module 302 supports dual-frequency transmission in the 2.4GHz and 5.8GHz frequency bands, and adaptively selects low-interference channels through a dynamic frequency band switching mechanism. The 2.4GHz frequency band provides a communication link with strong wall penetration capability and wide coverage, while the 5.8GHz frequency band uses its high bandwidth characteristics to ensure low packet loss rate transmission of high-definition video streams and multi-sensor data; the dual-frequency collaborative work uses frequency division multiplexing technology to improve communication stability in multipath reflection and electromagnetic interference scenarios in tunnels, and at the same time compresses the end-to-end communication delay to less than 50 milliseconds based on timestamp synchronization and priority scheduling algorithms, ensuring real-time issuance of operating instructions from the remote control terminal 6 and synchronous return of on-board perception data; the module's built-in anti-interference protocol and adaptive power regulation function further suppress signal attenuation caused by metal structures in the tunnel, ultimately achieving highly reliable and low-latency interaction of control instructions, environmental perception information and vehicle status data, and supporting seamless switching between manual remote control and autonomous driving modes.

[0051] The control system is built based on the edge computing architecture, and the onboard industrial computer 4 is used to fuse multi-sensor data and generate control decisions.

[0052] Specifically, the control system is built based on the edge computing architecture, and the parallel computing capability of the on-board industrial computer 4 is used to perform millisecond-level time synchronization and spatial alignment of multi-source heterogeneous data from the lidar 501, visual camera 502 and ultrasonic probe 503 to eliminate information conflicts between sensors; the lightweight deep learning model carried by the on-board industrial computer 4 analyzes environmental semantic information in real time, and at the same time integrates the posture data of the positioning module 301 and the global path constraints of the path planning algorithm unit 4011 to generate dynamic control instructions that take into account both obstacle avoidance efficiency and driving smoothness; the edge computing architecture reduces cloud dependence through localized data processing, and ensures the real-time performance of the full-link closed loop driven by multi-sensor data fusion, decision generation and execution module 402 in scenarios with restricted tunnel communication, ultimately achieving efficient coordination of the perception-decision-execution cycle during vehicle autonomous navigation, and supporting stable and reliable vehicle control response in complex dynamic environments.

[0053] The control module 401 includes a path planning algorithm unit 4011 and an obstacle avoidance unit 4012. The obstacle avoidance module implements dynamic path adjustment based on deep learning target recognition.

[0054] Specifically, the control module 401 includes a path planning algorithm unit 4011 and an obstacle avoidance unit 4012, wherein the path planning algorithm unit 4011 generates a global optimal path based on a global map and real-time positioning data, and at the same time combines the dynamic window method to smoothly optimize the local trajectory to ensure the vehicle's turning radius and executability under kinematic constraints; the obstacle avoidance unit 4012 uses a deep learning target recognition model to analyze the obstacle category, size and motion trend in the multimodal sensor data in real time, triggers a dynamic path adjustment strategy based on the safety distance threshold and predicted trajectory conflict detection, and uses an elastic time window algorithm to balance driving efficiency and safety during the obstacle avoidance process; the two realize the deep integration of static path planning and dynamic obstacle avoidance through a collaborative iterative mechanism, quickly generate collision-free paths in scenarios such as tunnel construction equipment movement and sudden intrusion of personnel, and improve the decision-making reliability under complex working conditions through closed-loop verification and feedback correction of control instructions, ultimately supporting the vehicle's fully autonomous navigation and adaptive obstacle avoidance capabilities in dynamic closed environments.

[0055] A lithium iron phosphate battery 7 is installed in the vehicle frame 1 , and a battery management unit 4013 is provided in the control module 401 for monitoring the battery status.

[0056] Specifically, a lithium iron phosphate battery 7 is installed in the frame 1, which provides continuous power supply for the vehicle through its high energy density and cycle stability, and adapts to the durability requirements of the high temperature, vibration and dust environment in the tunnel; the battery management unit 4013 integrated in the control module 401 collects the voltage, temperature and remaining power data of the battery pack in real time, optimizes the consistency of the battery cells and predicts the health status through a dynamic balancing algorithm, and intelligently adjusts the charging and discharging strategy in combination with the vehicle operating conditions to prevent the risks of overcharging, over-discharging and thermal runaway; the battery management unit 4013 controls the energy consumption of the entire vehicle in coordination with the on-board industrial computer 4 through a multi-level protection mechanism, while ensuring the stable operation of high-load equipment such as the multimodal sensor module 5 and the fusion positioning module 301, while extending the battery life, and ultimately achieving a dynamic balance between power system safety control and energy efficiency, supporting the vehicle's all-weather continuous operation capability in the complex environment of the tunnel.

[0057] The remote control terminal 6 includes a video monitoring unit 601 , an operation instruction unit 602 and a parameter trajectory echo unit 603 , and is used to realize human-computer interaction and remote control functions.

[0058] Specifically, the remote control terminal 6 includes a video monitoring unit 601, an operation instruction unit 602 and a parameter trajectory echo unit 603. The video monitoring unit 601 uses the laser point cloud, visual image and ultrasonic data stream collected by the multimodal sensor module 5 to render a three-dimensional visualization interface of the tunnel environment in real time through compression coding and low-latency transmission protocol, and simultaneously superimposes the vehicle posture information of the positioning module 301 to provide the operator with global situation awareness; the operation instruction unit 602 analyzes the operation instructions based on the human-computer interaction interface and generates a standardized control protocol, and ensures steering, addition and subtraction through redundancy check and priority scheduling mechanism. The reliable transmission of speed and emergency braking commands supports lossless switching between autonomous driving mode and manual control; the parameter trajectory echo unit 603 continuously records the vehicle's historical trajectory, sensor status and energy consumption data, and combines the timestamp synchronous playback function to achieve fault backtracking and behavior analysis, while triggering abnormal status warnings based on preset electronic fences and speed thresholds; the three form a closed-loop human-machine collaborative control through the interactive link of the wireless communication module 302, relying on the local caching strategy to maintain basic control functions in the event of tunnel network fluctuations or device offline scenarios, ultimately achieving transparent management of vehicle operation data throughout its life cycle and highly robust remote intervention capabilities.

[0059] Working principle: When the vehicle is driving in the tunnel, it realizes the three-dimensional modeling of the tunnel environment and the precise perception of the front and rear obstacles by deploying two 32-line laser radars 5011 (installed with a pitch angle of ±15°) at the front and rear of the vehicle body. At the same time, a blind spot-filling laser radar 5012 is arranged at the front and rear to cover the low area at close range and avoid blind spots. Multiple fisheye cameras are arranged at the four corners, top and both sides of the vehicle body to achieve 360° visual perception without blind spots, especially for the detection of lateral and far-end targets and blind spot compensation. Binocular cameras are installed in the front and rear cabs 2 to enhance the depth perception capability at both ends. Six ultrasonic probes 503 are integrated in the front and rear bumpers to realize wall-sticking driving assistance and low-speed collision avoidance warning. These sensor data are transmitted to the on-board industrial computer 4 in real time, and multi-source information fusion is completed through edge computing, and deep learning target detection algorithms (such as A YOLO-based model) identifies obstacle categories and relative positions, and then combines path planning algorithms (such as the improved A* algorithm) to achieve dynamic obstacle avoidance and route replanning. Vehicle positioning relies on a UWB and IMU fusion positioning system, which can provide high-precision location information in tunnel environments without GPS signals. At the same time, it cooperates with the 5.8GHz wireless communication module 302 to achieve low-latency data interaction with the remote control terminal 6 (transmission delay is less than 50ms), ensuring real-time command response. The execution layer includes driving, steering, and braking devices, and the control module 401 issues commands based on the calculation results to complete autonomous or remote control. The vehicle energy system uses a lithium iron phosphate battery 7 with a battery life of ≥8 hours, and is equipped with a battery management system (BMS) to achieve real-time monitoring of battery voltage, current, temperature, and remaining power to ensure the safety and sustainability of the entire vehicle operation. Through the efficient collaboration between the above systems, the smart rubber-wheeled vehicle can stably, efficiently, and intelligently complete material transportation tasks in complex tunnel environments.

[0060] This paper uses a deep learning-based target detection and path planning fusion algorithm to achieve intelligent obstacle avoidance. The specific workflow is as follows:

[0061] First, while driving, the vehicle uses the forward 32-line lidar 5011 and the front binocular camera to collaboratively perceive the environment ahead. The lidar 501 provides high-precision distance and spatial structure information, while the binocular camera collects images for semantic recognition. These perception data are synchronously input into an integrated deep learning model (such as YOLOv5) for target detection and classification, which can effectively identify typical obstacles in tunnels, including construction equipment, personnel, piles of materials, and wall protrusions.

[0062] When the recognition results determine that the target enters the vehicle's preset safety buffer zone (such as 5 meters in front and 1 meter to the side), the system triggers the obstacle avoidance decision process. At this time, the fused perception data will be input into the path planning module for path reconstruction, which includes two steps: first, building a safe navigation map based on the current map status and obstacle location; second, using the A* algorithm to generate a new optimal path, and further optimizing the trajectory through the dynamic window method (DWA) to ensure that the obstacle avoidance process is smooth, without sudden turns or stops, and meets the vehicle's turning radius constraints; in order to deal with dynamic obstacles (such as moving people or equipment), the system introduces the Kalman filter algorithm to predict the target's future motion trajectory, and combines the time window to determine whether it will pose a threat to the current travel path. If it is judged to be a high-risk dynamic interference, the system will prioritize deceleration or short-term parking, and automatically resume driving after the obstacle is eliminated or the path is cleared.

[0063] Ultimately, the path planning results are converted into control instructions and sent by the control module 401 to the vehicle's steering mechanism, drive system, and braking unit to complete the actual obstacle avoidance action. The entire process is completed in real time by the on-board edge computing unit. The obstacle avoidance response delay is less than 200ms, and a manual takeover mechanism is supported to ensure operational safety under complex working conditions.

[0064] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent rubber-wheeled vehicle for use in tunnels, comprising a vehicle frame (1), characterized in that: Two cabs (2) are fixedly connected to both ends of the vehicle frame (1), and a gateway switch (3) and an on-board industrial control computer (4) are fixedly connected to the upper surface of the cab (2); A multimodal sensor module (5) is provided outside the cab (2), and the multimodal sensor module (5) includes a laser radar (501), a visual camera (502), and an ultrasonic probe (503), and is used to sense the environment in the tunnel; The gateway switch (3) is internally provided with a fusion positioning module (301) and a wireless communication module (302) for realizing vehicle positioning and remote communication; The vehicle-mounted industrial control computer (4) is internally provided with a control module (401) and an execution module (402), which are used to receive sensing information and generate control instructions, and then drive the vehicle to operate according to the control instructions; A remote control terminal (6) is provided inside the cab (2) for monitoring the vehicle status and realizing remote operation.

2. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The laser radar (501) includes four 32-line laser radars (5011) and two blind spot laser radars (5012), wherein the four 32-line laser radars (5011) are respectively installed at the rearview mirrors of the two cabs (2), and the two blind spot laser radars (5012) are installed at the middle of the front ends of the two cabs (2).

3. The intelligent rubber-tyred vehicle for use in tunnels according to claim 2, characterized in that: The visual camera (502) includes two binocular cameras (5021) and fourteen 360 fisheye cameras (5022), the two binocular cameras (5021) are installed at the middle of the front end of the two cabs (2), the binocular camera (5021) is located directly above the blind spot laser radar (5012), and the fourteen 360 fisheye cameras (5022) are respectively installed at the upper side of the front end of the two cabs (2), at the four corners of the front end of the two cabs (2), and at the upper end of the side surfaces of the two cabs (2).

4. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The ultrasonic probes (503) are installed at the bumpers of the two cabs (2), and twelve are installed in total. The detection distance of the ultrasonic probes (503) is 0.1-3m, and the accuracy is ±2cm.

5. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The fusion positioning module (301) comprises a UWB positioning unit (3011) and an IMU inertial navigation unit (3012), and the fusion positioning module (301) achieves high-precision positioning through data fusion.

6. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The wireless communication module (302) supports dual-frequency transmission in the 2.4 GHz and 5.8 GHz frequency bands, and the communication delay is less than 50 milliseconds.

7. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The control system is constructed based on an edge computing architecture, and the onboard industrial computer (4) is used to fuse multi-sensor data and generate control decisions.

8. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The control module (401) comprises a path planning algorithm unit (4011) and an obstacle avoidance unit (4012), wherein the obstacle avoidance module implements dynamic path adjustment based on deep learning target recognition.

9. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: A lithium iron phosphate battery (7) is installed in the vehicle frame (1), and a battery management unit (4013) is provided in the control module (401) for monitoring the battery status.

10. The intelligent rubber-tyred vehicle for use in tunnels according to claim 1, characterized in that: The remote control terminal (6) comprises a video monitoring unit (601), an operation instruction unit (602) and a parameter trajectory echo unit (603), and is used to realize human-computer interaction and remote control functions.

Images