Vehicle front exit accompanying cable control type ground defect detection system

Abstract

The invention discloses a vehicle front-out accompanying cable control type ground defect detection system. The system comprises one or more motion detection mechanisms which are connected with a vehicle through composite cables. The motion detection mechanism is provided with an ice-penetrating / ground-penetrating radar, a laser radar, a detection / navigation camera and other sensors and a pose module, and sensor data is collected through a data interface module and then sent to a vehicle in real time through a composite cable. After the vehicle receives the data of the sensor, the vehicle-mounted intelligent sensing control unit processes the data to obtain a ground defect detection result and an environment sensing result; and according to a ground defect detection result, an environment perception result and vehicle control information, a driving path, a driving speed and a control command of the motion detection mechanism are calculated and are sent to the motion detection mechanism in real time through the composite cable for execution. Advanced perception of the front ground defect by the vehicle is realized in a forward perception and backward calculation mode, and the driving safety of the vehicle is ensured in real time.

Description

Technical Field

[0001] The present application relates to the field of unmanned intelligent driving technology, and in particular to a vehicle-mounted cable-controlled ground defect detection system. Background Art

[0002] Intelligent driving and unmanned driving are at the forefront of today's high-tech development, and have already brought about and will bring about a huge release of productivity. Civilian unmanned vehicles generally travel on high-grade roads, without considering obstacles below the road surface. However, for engineering machinery and special ground unmanned vehicles traveling under complex geological conditions such as plateaus, high cold, polar regions, and ice surfaces, ground defects such as pits, traps, collapses, puddles, and ice cracks seriously affect their traffic capacity and cause them to be overturned, fall, and other major dangers, and a safe and reliable detection method is urgently needed.

[0003] Environmental perception, as an important part of intelligent unmanned driving, has made a lot of achievements. For example, in the field of civilian unmanned driving, there are pure image vision solutions that only use cameras for some vehicle driving, or multi-source sensor solutions of laser radar and cameras for other unmanned driving. However, there is little in-depth research on environmental perception below the road surface. Part of the reason is that environmental perception below the road surface is very difficult. For example, when a vehicle is driving at high speed and meets the need for vehicle braking, there is no on-board sensor that can detect the road surface that is about to collapse 20 meters in front of the vehicle. Although ice-penetrating radar and ground-penetrating radar can detect road collapse that is about to collapse, they all require downward installation, that is, they must be installed under the vehicle body, but installing them under the vehicle body cannot meet the braking distance requirements, and the method of mounting them at the rear of the vehicle cannot detect ahead of time and cannot meet the needs of the vehicle moving forward.

[0004] The patent "An Ice Structure Detection Robot and Detection Method (Patent Application No. 202010493297.5)" provides an ice structure detection robot, including a mobile robot, a visible ice crack and obstacle detection system, a radar detection system, a snow sample collection system and a robot-side controller, which are used to solve the problems of detecting large-scale glaciers, determining safe routes on ice, and automatically collecting glacier snow samples. It is mainly aimed at application scenarios such as pre-exploration of safe routes and geological surveys. Although the above patent provides a robot that can be used for ice surface detection to detect invisible ice crack information and ice structure information, it has the following problems:

[0005] 1. The robot is self-sufficient in energy and power. The battery, engine, etc. are placed on the robot body, which increases the weight of the robot and thus increases the risk of falling when driving on the ice.

[0006] 2. The robot processes sensor data on its main body. The robot is equipped with a visible ice crack and obstacle detection system, a radar detection system, and a snow sample collection system. These systems on the robot process data from sensors such as ice-penetrating radar, lidar, and cameras, further increasing the weight and cost of the robot.

[0007] 3. The robot's route planning and obstacle avoidance control are completed by the controller on the robot body (robot-side control). Since the robot is small and short, its field of view is limited, and the efficiency of route planning and obstacle avoidance will be low.

[0008] 4. The robot does not have fall protection and post-fall disposal devices. It communicates wirelessly with the operator (monitoring computer) and has no mechanical connection. Once it encounters thin ice, it is extremely easy to fall into the ice cracks and cannot automatically handle the fall. Summary of the invention

[0009] The present application provides a vehicle-front cable-controlled ground defect detection system, which can be used to solve the problem of advanced perception of the ground ahead during vehicle driving and achieve safe passage of the vehicle in dangerous environments.

[0010] A vehicle-mounted cable-controlled ground defect detection system is characterized in that the system includes one or more motion detection mechanisms, a composite cable for connecting the motion detection mechanism and the vehicle, a cable-releasing and -retracting unit, a power transmission unit, and an on-board intelligent sensing control unit.

[0011] The motion detection mechanism includes ice-penetrating / ground-penetrating radar, lidar, perception / navigation camera, positioning and attitude determination unit, intelligent data interface unit, and walking mechanism.

[0012] The ice-penetrating / ground-penetrating radar is installed at the bottom of the motion detection mechanism, facing vertically downward; the laser radar and perception / navigation camera are installed at the front of the motion detection mechanism, facing forward; the positioning and attitude unit is installed at the upper center of the vehicle, tightly hard-linked to the vehicle body; the intelligent data interface unit is installed in the rear area of ​​the vehicle.

[0013] The motion detection mechanisms are respectively arranged at preset distances in front of the front traveling mechanisms of the vehicle, and are used to respectively detect the areas in front of each traveling mechanism.

[0014] Further, the motion detection mechanisms are each connected to the vehicle via a composite cable; the power transmission unit on the vehicle supplies power to the motion detection mechanisms via the composite cable;

[0015] The motion detection mechanism detects forward, and the sensor data loaded by the motion detection mechanism is collected by the intelligent data interface unit and then transmitted to the vehicle through the composite cable;

[0016] After the vehicle receives the data, it is processed by the on-board intelligent perception control unit to obtain the detection results of ground defects and generate environmental perception results;

[0017] The vehicle's on-board intelligent perception control unit integrates the detection results, environmental perception results and the vehicle's current status information, processes the driving path, driving speed and control commands of the motion detection mechanism, and sends them to the motion detection mechanism through a composite cable.

[0018] Furthermore, when the sensor data carried by the motion detection mechanism are collected through the intelligent data interface unit, the system clock of the motion control mechanism is used as a unified clock to obtain the ground point cloud data of the ice-penetrating / ground-penetrating radar, the surrounding environment point cloud data of the lidar, and the perception / navigation camera image data.

[0019] The intelligent data interface unit aligns and packages the sensor data of ice-penetrating / ground-penetrating radar, lidar, perception / navigation camera, and positioning and attitude unit, and assigns a timestamp, and sends the data in real time to the on-board intelligent perception control unit on the vehicle via a composite cable.

[0020] Furthermore, the vehicle-mounted intelligent perception control unit integrates the detection results and the environmental perception results as well as the vehicle's current status information, including:

[0021] The on-board intelligent perception control unit on the vehicle generates local environment point cloud data by fusing the multi-frame point cloud data obtained by ice-penetrating / ground-penetrating radar and lidar based on the position and posture information of the motion detection mechanism. data Mapping to the feature space to get F vox ,

[0022] F vox =voxel(F data )

[0023] Among them, voxel means voxelization;

[0024] Extract three point cloud features F of different scales respectively stage1 、F stage2 、F stage3 :

[0025] F stage1 =ReLU(3DConv(F vox ))

[0026] F stage2 =ReLU(3DConv(F stage1 ))

[0027] F stage3 =ReLU(3DConv(F stage2 ))

[0028] Among them, 3DConv is a 3D convolution operation, and ReLU is an activation function;

[0029] At the same time, the texture features of images of different scales are extracted at three scales. stage1 ,I stage2 ,I stage3 :

[0030] I stage1 =ReLU(Conv(I))

[0031] I stage2 =ReLU(Conv(I stage1 ))

[0032] I stage3 =ReLU(Conv(I stage2 ))

[0033] Among them, Conv is the convolution operation and ReLU is the activation function;

[0034] Fusion of point cloud and image multimodal features,

[0035] F 1 =conv(concat(F stage1 +I stage1 ))

[0036] F 2 =conv(concat(F stage2 +I stage2 ))

[0037] F 3 =conv(concat(F stage3 +I stage3 ))

[0038]

[0039] Among them, concat is a channel-by-channel concatenation operation. It is a bitwise accumulation operation;

[0040] Realize real-time detection of ground classes such as pits, traps, collapses, puddles, and ice cracks:

[0041] class = softmax(MLP(F))

[0042] Among them, MLP is a multi-layer perceptron and softmax is an activation function.

[0043] Furthermore, the vehicle-mounted intelligent perception control unit (5) uses an LSTM temporal feature extraction network to extract the jump features of the positioning and posture determination unit (9):

[0044] F imu =LSTM(X k )

[0045] Among them, X k is the current motion detection mechanism (1) posture information, LSTM is the temporal feature extraction network;

[0046] Using the jump characteristics, the fall detection of the motion detection mechanism (1) is realized:

[0047] O = softmax(F imu ).

[0048] Furthermore, the on-board intelligent perception control unit is also used to analyze the sensor data of the motion detection mechanism, use lidar and perception cameras to detect the vehicle's surrounding environment in real time, mark obstacle information, and generate local environmental obstacle results based on the position and posture information of the motion detection mechanism. At the same time, it combines ground defect detection to improve the ground defect information of the local environmental perception results and generate local environmental perception results.

[0049] Furthermore, the vehicle-mounted intelligent perception control unit performs path planning, speed planning and control of the motion detection mechanism, specifically including:

[0050] According to the local environment perception results and the position and posture information of the motion detection mechanism, the travel path of each motion detection mechanism is planned to avoid obstacle areas and ground defect areas, and the travel trajectories of the current vehicle and multiple motion detection mechanisms are estimated at the same time;

[0051] The distance between the motion detection mechanism and the vehicle is set to D, and the motion detection mechanism and the vehicle are controlled to maintain a distance before and after the error is no greater than Initial planning speed v of the motion detection mechanism 1 Keeping in line with the speed of the vehicle, the actual distance D between the motion detection mechanism and the vehicle is 1 The difference between the set distance D and the final speed v is

[0052]

[0053] Among them, min is the minimum value operation.

[0054] Further, it is applicable to two types of vehicles, one is a manned vehicle and the other is an unmanned vehicle:

[0055] When a manned vehicle is used, a display control unit is configured in the vehicle cockpit. The vehicle-mounted intelligent perception control unit uses the advance perception results of the motion detection mechanism to generate the perception results of the terrain obstacles in front of the vehicle and displays them on the display control unit to assist the driver in understanding the road conditions. When there are dangerous obstacles on the vehicle's route, the vehicle brakes and reminds the driver;

[0056] When using unmanned vehicles, they can achieve advanced perception through motion detection mechanisms, allowing safe passage in complex environments.

[0057] Furthermore, the cable releasing and collecting unit includes a cable storage winch, a traction winch, an automatic cable arranger and a slack compensator; and is responsible for releasing and collecting the composite cable.

[0058] The friction wheel on the traction winch and the cable storage device of the cable storage winch are both equipped with a servo motor with an encoder and a reducer with a suitable reduction ratio;

[0059] When the cable needs to be retracted or released, the two friction wheels in the traction winch rotate in the same direction, and the cable storage device of the cable storage winch cooperates with it to realize the retraction and release of the cable. At the same time, the rotation direction is determined according to the encoder on the motor, and then the retraction or release of the cable is judged and controlled;

[0060] When the cable is extended to the required length, it is fixed. On the one hand, the cable storage winch is mechanically limited. On the other hand, the friction wheel on the left side of the traction winch rotates slowly clockwise, and the friction wheel on the right side rotates slowly counterclockwise. At this time, the cable is subjected to tension in two directions. At the same time, the stress and strain of the cable at the tensioning point are detected by the strain sensor. When the threshold required for tensioning is reached, the motor stops rotating and is guaranteed not to reverse through the ratchet on the transmission rod, thereby achieving a secure fixation of the cable.

[0061] The present invention proposes a new ground defect detection system of "pre-exit accompanying, pre-sensing and post-calculation". Compared with the prior art, the present invention has the following significant advantages:

[0062] 1. The motion detection mechanism accompanying the front realizes the advanced environmental perception of the moving vehicle;

[0063] 2. The environmental perception sensors on the motion detection mechanism only complete the data collection of ground defects such as pits, traps, collapses, puddles, ice cracks, etc., but the data processing is handled by the vehicle's on-board intelligent perception control unit, thereby reducing the weight, cost and complexity of the motion detection mechanism;

[0064] 3. The motion detection mechanism obtains power from the vehicle, with strong protection capability, and because it does not carry batteries or oil engines, it is light in weight and not easy to fall;

[0065] 4. The data transmission between the motion detection mechanism and the vehicle adopts the wired communication method of composite cable, which has the advantages of no delay, large bandwidth and anti-interference;

[0066] 5. The motion detection mechanism and the vehicle are connected by a composite cable, which has the advantages of fall protection and easy disposal after fall. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0068] Figure 1 is a system structure diagram of the present invention;

[0069] Figure 2 It is a schematic diagram of the structure of the motion detection mechanism of the present invention;

[0070] Figure 3 The figure is a schematic structural diagram of a cable releasing and receiving unit according to an embodiment of the present invention.

[0071] Description of reference numerals:

[0072] 1-motion detection mechanism; 2-composite cable; 3-cable releasing and collecting unit; 4-power transmission unit; 5-vehicle intelligent sensing control unit;

[0073] 6- ice-penetrating / ground-penetrating radar; 7- lidar; 8- perception / navigation camera; 9- positioning and attitude determination unit; 10- intelligent data interface unit; 11- walking mechanism;

[0074] 12-cable storage winch; 13-traction winch; 14-automatic cable arranger; 15-slack compensator. DETAILED DESCRIPTION

[0075] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, a person skilled in the art can imagine various embodiments of the present invention. Therefore, the following specific embodiments and drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as the whole of the present invention or as a limitation or limitation of the technical solution of the present invention. On the contrary, the purpose of providing these embodiments is to enable those skilled in the art to understand the present invention more thoroughly. The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the innovative concept of the present invention.

[0076] This embodiment includes two motion detection mechanisms 1, a composite cable 2 connecting the motion detection mechanisms and the vehicle, a cable releasing and retracting unit 3, a power transmission unit 4, and an on-board intelligent sensing control unit 5.

[0077] The motion detection mechanism 1 includes an ice-penetrating / ground-penetrating radar 6, a laser radar 7, a perception / navigation camera 8, a positioning and posture determination unit 9, an intelligent data interface unit 10, and a walking mechanism 11.

[0078] Among them, the ice-penetrating / ground-penetrating radar 6 is installed at the bottom of the motion detection mechanism 1, facing vertically downward; the laser radar 7 and the perception / navigation camera 8 are installed at the front of the motion detection mechanism 1, facing forward; the positioning and attitude unit 9 is installed at the upper center of the vehicle, tightly hard-linked to the vehicle body; the intelligent data interface unit 10 is installed in the rear area of ​​the vehicle.

[0079] In this embodiment, two motion detection mechanisms are used, each arranged 25 meters in front of the front running mechanism of the vehicle, to detect the 25-meter area in front of each running mechanism. The two motion detection mechanisms 1 are each connected to the vehicle via a composite cable 2; the power transmission unit 4 on the vehicle supplies power to the motion detection mechanism 1 via the composite cable 2;

[0080] The motion detection mechanism 1 detects forward, and the sensor data loaded by the motion detection mechanism 1 is collected through the intelligent data interface unit 10 and transmitted to the vehicle through the composite cable 2; after the vehicle receives the data, it is processed by the on-board intelligent perception control unit 5 to obtain the detection results of ground defects and generate environmental perception results. The on-board intelligent perception control unit 5 on the vehicle processes the driving path, driving speed and control command of the motion detection mechanism 1 according to the detection results, environmental perception results and the current status information of the vehicle, and sends them to the motion detection mechanism 1 through the composite cable 2.

[0081] When the sensor data loaded by the motion detection mechanism 1 is collected through the intelligent data interface unit 10, the system clock of the motion control mechanism 1 is used as a unified clock to obtain the ground point cloud data of the ice-penetrating / ground-penetrating radar 6, the surrounding environment point cloud data of the lidar 7, and the image data of the perception / navigation camera 8. After the data is packaged, a timestamp is assigned and the data is sent in real time to the on-board intelligent perception control unit 5 on the vehicle via the composite cable 2.

[0082] The vehicle-mounted intelligent perception control unit 5 on the vehicle generates local environment point cloud data by fusing the multi-frame point cloud data obtained by the ice-penetrating / ground-penetrating radar and the laser radar based on the 1-bit posture information of the motion detection mechanism. data Mapping to the feature space to get F vox ,

[0083] F vox =voxel(F data )

[0084] Among them, voxel means voxelization;

[0085] Extract three point cloud features F of different scales respectively stage1 、Fstage2 、F stage3 :

[0086] F stage1 =ReLU(3DConv(F vox ))

[0087] F stage2 =ReLU(3DConv(F stage1 ))

[0088] F stage3 =ReLU(3DConv(F stage2 ))

[0089] Among them, 3DConv is a 3D convolution operation, and ReLU is an activation function;

[0090] At the same time, the texture features of images of different scales are extracted at three scales. stage1 ,I stage2 ,I stage3 :

[0091] I stage1 =ReLU(Conv(I))

[0092] I stage2 =ReLU(Conv(I stage1 ))

[0093] I stage3 =ReLU(Conv(I stage2 ))

[0094] Among them, Conv is the convolution operation and ReLU is the activation function;

[0095] Fusion of point cloud and image multimodal features,

[0096] F 1 =conv(concat(F stage1 +I stage1 ))

[0097] F 2 =conv(concat(F stage2 +I stage2 ))

[0098] F 3 =conv(concat(F stage3 +I stage3 ))

[0099]

[0100] Among them, concat is a channel-by-channel concatenation operation. It is a bitwise accumulation operation;

[0101] Realize real-time detection of ground classes such as pits, traps, collapses, puddles, and ice cracks:

[0102] class = softmax(MLP(F))

[0103] Among them, MLP is a multi-layer perceptron and softmax is an activation function.

[0104] The vehicle-mounted intelligent perception control unit 5 uses the LSTM time series feature extraction network to extract the jump features of the positioning and posture determination unit 9:

[0105] F imu =LSTM(X k )

[0106] Among them, X k is the current motion detection mechanism 1 posture information, and LSTM is the temporal feature extraction network;

[0107] Using the jump characteristics, the fall detection of the motion detection mechanism 1 is realized:

[0108] O = softmax(F imu ).

[0109] The on-board intelligent perception control unit 5 is also used to analyze the sensor data of the motion detection mechanism 1, use the laser radar and perception camera to detect the vehicle's surrounding environment in real time, mark obstacle information, generate local environmental obstacle results based on the position and posture information of the motion detection mechanism 1, and at the same time combine ground defect detection to improve the ground defect information of the local environmental perception results to generate local environmental perception results.

[0110] The vehicle-mounted intelligent perception control unit plans the travel paths of the two motion detection mechanisms 1 according to the local environment perception results and the posture information of the motion detection mechanism 1 to avoid the obstacle area and the ground defect area, and estimates the travel trajectory of the current vehicle and the two motion detection mechanisms 1. In this embodiment, the distance between the motion detection mechanism 1 and the vehicle is set to 25 meters, and the motion detection mechanism 1 and the vehicle are controlled to maintain an error of no more than 2.5 meters before and after the set distance. The initial planned speed of the motion detection mechanism 1 is 10km / h, which is consistent with the speed of the vehicle. According to the difference between the actual distance 20 between the motion detection mechanism (1) and the vehicle and the set distance 25 meters, the final speed v is

[0111]

[0112] Among them, min is the minimum value operation.

[0113] There are two types of vehicles to which this patent applies, one is a manned vehicle and the other is an unmanned vehicle: when a manned vehicle is used, a display control unit is configured in the vehicle cockpit, and the on-board intelligent perception control unit 5 uses the advance perception results of the motion detection mechanism 1 to generate a terrain obstacle perception result in front of the vehicle and displays it on the display control unit, thereby assisting the driver in understanding the road conditions and, when there are dangerous obstacles on the vehicle's route, controlling the vehicle to brake and reminding the driver; when an unmanned vehicle is used, the unmanned vehicle achieves advance perception through the motion detection mechanism 1, thereby achieving safe passage in complex environments.

[0114] The cable releasing and collecting unit in the present invention comprises a cable storage winch 12 , a traction winch 13 , an automatic cable arranger 14 and a slack compensator 15 ; and is responsible for releasing and collecting the composite cable 2 . The friction wheel on the traction winch 13 and the cable storage device of the cable storage winch 12 are both equipped with a servo motor with an encoder and a reducer with a suitable reduction ratio; when the cable needs to be retracted or released, the two friction wheels in the traction winch 13 rotate in the same direction, and the cable storage device of the cable storage winch 12 cooperates with it to realize the retraction and release of the cable, and at the same time determines the rotation direction according to the encoder on the motor, and then judges and controls the retraction or release of the cable; when the cable is extended to a required length, it is fixed. On the one hand, the cable storage winch 12 is mechanically limited, and on the other hand, the friction wheel on the left side of the traction winch 13 rotates slowly clockwise, and the friction wheel on the right side rotates slowly counterclockwise. At this time, the cable is subjected to tensioning forces in two directions; at the same time, the stress and strain of the cable at the tensioning point are detected by a strain sensor. After reaching the threshold required for tensioning, the motor stops rotating and is guaranteed not to reverse through the ratchet on the transmission rod, thereby achieving a secure fixation of the cable.

[0115] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

[0116] It should be understood that in order to simplify the present invention and help those skilled in the art understand the various aspects of the present invention, in the above description of the exemplary embodiments of the present invention, the various features of the present invention are sometimes described in a single embodiment or described with reference to a single figure. However, the present invention should not be interpreted as the features included in the exemplary embodiments are all necessary technical features of the patent claims.

[0117] It should be understood that the modules, units, components, etc. included in the device of an embodiment of the present invention can be adaptively changed to be arranged in a device different from the embodiment. The different modules, units, or components included in the device of the embodiment can be arranged into one module, unit, or component, or they can be divided into multiple sub-modules, sub-units, or sub-components.

Claims

1. A vehicle-mounted cable-controlled ground defect detection system, characterized in that: The system comprises one or more motion detection mechanisms (1), a composite cable (2) for connecting the motion detection mechanism (1) and the vehicle, a cable releasing and collecting unit (3), a power transmission unit (4), and an on-board intelligent sensing control unit (5); The motion detection mechanism (1) includes an ice-penetrating / ground-penetrating radar (6), a laser radar (7), a perception / navigation camera (8), a positioning and posture determination unit (9), an intelligent data interface unit (10), and a walking mechanism (11); The ice-penetrating / ground-penetrating radar (6) is installed at the bottom of the motion detection mechanism (1) in a vertically downward direction; the laser radar (7) and the perception / navigation camera (8) are installed at the front of the motion detection mechanism (1) in a forward direction; the positioning and posture determination unit (9) is installed at the center of the upper part of the vehicle and is hard-linked to the vehicle body; the intelligent data interface unit (10) is installed in the rear area of ​​the vehicle; The motion detection mechanisms (1) are respectively arranged at a preset distance in front of the front running mechanisms of the vehicle and are used to respectively detect the areas in front of each running mechanism.

2. The vehicle-front cable-controlled ground defect detection system according to claim 1 is characterized in that: The motion detection mechanisms (1) are each connected to a vehicle via a composite cable (2); a power transmission unit (4) on the vehicle supplies power to the motion detection mechanisms (1) via the composite cable (2); The motion detection mechanism (1) moves forward for detection, and sensor data loaded on the motion detection mechanism (1) is collected through an intelligent data interface unit (10) and then transmitted to the vehicle through a composite cable (2); After the vehicle receives the data, the data is processed by the vehicle-mounted intelligent sensing control unit (5) to obtain the detection result of the ground defect and generate the environmental sensing result; The vehicle-mounted intelligent sensing control unit (5) fuses the detection results with the environmental sensing results and the current state information of the vehicle, processes and obtains the driving path, driving speed and control command of the motion detection mechanism (1), and sends them to the motion detection mechanism (1) via the composite cable (2).

3. The vehicle-front accompanying cable-controlled ground defect detection system according to claim 2 is characterized in that: When the sensor data loaded by the motion detection mechanism (1) is collected through the intelligent data interface unit (10), the system clock of the motion control mechanism (1) is used as a unified clock to obtain the ground point cloud data of the ice-penetrating / ground-penetrating radar (6), the surrounding environment point cloud data of the laser radar (7), and the image data of the perception / navigation camera (8); The intelligent data interface unit (10) completes the alignment and packaging of the sensor data of the ice-penetrating / ground-penetrating radar (6), the laser radar (7), the perception / navigation camera (8), and the positioning and attitude unit (9), and assigns a time stamp, and sends the data to the on-board intelligent perception control unit (5) on the vehicle in real time via the composite cable (2).

4. The vehicle-front cable-controlled ground defect detection system according to claim 2 is characterized in that: The vehicle-mounted intelligent sensing control unit (5) integrates the detection results, the environmental sensing results and the current status information of the vehicle, including: The vehicle-mounted intelligent sensing control unit (5) on the vehicle generates local environment point cloud data by fusing the multi-frame point cloud data obtained by the ice-penetrating / ground-penetrating radar and the laser radar according to the position and posture information of the motion detection mechanism (1). data Mapping to the feature space to get F vox , F vox =voxel(F data ) Among them, voxel means voxelization; Extract three point cloud features F of different scales respectively stage1 、F stage2 、F stage3 : F stage1 =ReLU(3DConv(F vox )) F stage2 =ReLU(3DConv(F stage1 )) F stage3 =ReLU(3DConv(F stage2 )) Among them, 3DConv is a 3D convolution operation, and ReLU is an activation function; At the same time, the texture features of images of different scales are extracted at three scales. stage1 ,I stage2 ,I stage3 : and stage1 =ReLU(Conv(I)) and stage2 =ReLU(Conv(I stage1 )) and stage3 =ReLU(Conv(I stage2 )) Among them, Conv is the convolution operation and ReLU is the activation function; Fusion of point cloud and image multimodal features, F1=conv(concat(F stage1 +I stage1 )) F2=conv(concat(F stage2 +I stage2 )) F3=conv(concat(F stage3 +I stage3 )) Among them, concat is a channel-by-channel concatenation operation. It is a bitwise accumulation operation; Realize real-time detection of ground classes such as pits, traps, collapses, puddles, and ice cracks: class = softmax(MLP(F)) Among them, MLP is a multi-layer perceptron and softmax is an activation function.

5. The vehicle-front accompanying cable-controlled ground defect detection system according to claim 4 is characterized in that: The vehicle-mounted intelligent perception control unit (5) uses the LSTM time series feature extraction network to extract the jump features of the positioning and posture determination unit (9): F imu =LSTM(X k ) Among them, X k is the current motion detection mechanism (1) posture information, LSTM is the temporal feature extraction network; Using the jump characteristics, the fall detection of the motion detection mechanism (1) is realized: O=softmax(F imu )。 6. The vehicle-front accompanying cable-controlled ground defect detection system according to claim 2 is characterized in that: The vehicle-mounted intelligent perception control unit (5) is also used to analyze the sensor data of the motion detection mechanism (1), use the laser radar and the perception camera to detect the vehicle's surrounding environment in real time, mark obstacle information, generate local environmental obstacle results based on the position and posture information of the motion detection mechanism (1), and at the same time combine the ground defect detection to improve the ground defect information of the local environmental perception result, and generate a local environmental perception result.

7. The vehicle-front accompanying cable-controlled ground defect detection system according to claim 2 is characterized in that: The vehicle-mounted intelligent sensing control unit (5) performs path planning, speed planning and control of the motion detection mechanism (1), specifically including: Planning the travel path of each motion detection mechanism (1) according to the local environment perception result and the position and posture information of the motion detection mechanism (1) to avoid obstacle areas and ground defect areas, and at the same time estimating the travel trajectories of the current vehicle and multiple motion detection mechanisms (1); The distance between the motion detection mechanism (1) and the vehicle is set to D, and the motion detection mechanism (1) and the vehicle are controlled to maintain a distance before and after the set distance with an error of no more than The initial planned speed v1 of the motion detection mechanism (1) is consistent with the speed of the vehicle. The final speed v is adjusted according to the difference between the actual distance D1 between the motion detection mechanism (1) and the vehicle and the set distance D. Among them, min is the minimum value operation.

8. The vehicle-front accompanying cable-controlled ground defect detection system according to claim 3 is characterized in that: There are two types of vehicles: one is a manned vehicle and the other is an unmanned vehicle; When a manned vehicle is used, a display control unit is configured in the vehicle cockpit, and the vehicle-mounted intelligent perception control unit (5) generates a terrain obstacle perception result in front of the vehicle using the advance perception result of the motion detection mechanism (1) and displays it on the display control unit to assist the driver in understanding the road conditions. When there is a dangerous obstacle on the vehicle's route, the vehicle brakes and reminds the driver; When the unmanned vehicle is used, the unmanned vehicle achieves advanced perception through a motion detection mechanism (1).

9. The vehicle-front accompanying cable-controlled ground defect detection system according to claim 3 is characterized in that: The cable releasing and collecting unit comprises a cable storage winch (12), a traction winch (13), an automatic cable arranger (14) and a slack compensator (15); and is responsible for releasing and collecting the composite cable (2); The friction wheel on the traction winch (13) and the cable storage device of the cable storage winch (12) are both equipped with a servo motor with an encoder and a reducer with a suitable reduction ratio; When the cable needs to be retracted or released, the two friction wheels in the traction winch (13) rotate in the same direction, and the cable storage device of the cable storage winch (12) cooperates with it to realize the retraction and release of the cable. At the same time, the rotation direction is determined according to the encoder on the motor, and then the retraction or release of the cable is judged and controlled; When the cable is extended to a required length, it is fixed. On the one hand, the cable storage winch (12) performs mechanical limiting. On the other hand, the left friction wheel of the traction winch (13) rotates slowly clockwise, and the right friction wheel rotates slowly counterclockwise. At this time, the cable is subjected to tension in two directions. At the same time, the stress and strain of the cable at the tensioning point are detected by the strain sensor. When the threshold required for tensioning is reached, the motor stops rotating and is guaranteed not to reverse through the ratchet on the transmission rod, thereby achieving a secure fixation of the cable.

Images