TBM tunnel electric double-head transport vehicle has manned and unmanned double-mode control system

Abstract

The application discloses a TBM tunnel electric double-head transport vehicle manned and unmanned dual-mode control system, belongs to the field of special transport control of tunnel construction, and realizes high-precision positioning and full-working-condition environment sensing by constructing a multi-level control system architecture, and constructs a progressive decision chain of "environment sensing mark, dynamic safety envelope calculation, state evaluation matrix priority decision", realizes adaptive optimal switching of manned mode and unmanned mode based on real-time transport working conditions, and simultaneously deeply couples the double-head vehicle body architecture and the dual-mode control system, activates the corresponding end of the advancing direction and locks the rear end when the decision is the manned mode, eliminates the double-end control conflict, locks both ends of the driving operation mechanism when the decision is the unmanned mode, eliminates the risk of accidental collision, realizes all-directional environment sensing coverage and flexible bidirectional driving of the vehicle in the long and narrow space of the TBM tunnel, and significantly improves the maneuverability, safety and intelligent level of tunnel transportation.

Description

Technical Field

[0001] This invention relates to the field of special transportation control in tunnel construction, and more specifically, to a dual-mode control system for manned and unmanned TBM tunnel electric dual-head transport vehicles. Background Technology

[0002] With the continuous advancement of infrastructure construction and underground space development, the application of full-face tunnel boring machines (TBMs) is becoming increasingly widespread. During TBM tunnel construction, it is necessary to frequently transport tunnel segments, excavated soil, construction materials, and large equipment. However, the narrow space, rugged road surface, and limited ventilation inside the tunnel place extremely high demands on the mobility, safety, and environmental adaptability of the transport equipment.

[0003] Conventional tunnel transport vehicles mostly only support manual driving and lack autonomous driving capabilities. In the special scenario of tunnels, long-distance repetitive transportation can easily lead to driver fatigue, and the presence of harsh environments such as dust and harmful gases in certain transport sections within the tunnel can also pose a threat to personnel health.

[0004] Currently, some existing technologies propose the concept of dual driving modes. For example, patent publication number CN108334072A proposes a dual driving mode control system for sweepers based on Beidou navigation. This solution achieves switching between manned and unmanned driving through a mode switching switch. However, its mode switching relies solely on manual operation of the switch or brake pedal, making it mainly suitable for sweeping operations on open roads. Furthermore, the perception scheme of this patent also relies entirely on satellite wireless signals from the Beidou satellite navigation system for positioning. In enclosed and sheltered environments such as tunnels, satellite signals are blocked by surrounding rocks and linings, and the receiver cannot properly capture satellite signals. In other words, it lacks autonomous decision-making capabilities based on environmental perception and working condition assessment, making it difficult to adaptively select the optimal driving mode according to the dynamic changes in the transportation scenario.

[0005] Therefore, in response to the actual shortcomings, a dual-mode control system for manned and unmanned driving of the TBM tunnel electric dual-head transport vehicle is proposed. Summary of the Invention

[0006] The purpose of this invention is to solve the above-mentioned problems. A dual-mode control system for manned and unmanned driving of TBM tunnel electric dual-head transport vehicle is provided.

[0007] The objective of this invention can be achieved through the following technical solution: a dual-mode control system for manned and unmanned driving of a TBM tunnel electric dual-head transport vehicle, applied to a dual-head transport vehicle with a dual-head body structure, comprising:

[0008] The perception layer is used to collect perception data of the target tunnel transportation scenario and motion status data of the dual-headed transport vehicle. The perception data includes positioning observation data and environmental observation data.

[0009] The communication layer is used to establish a data link between the dual-head transport vehicle and the ground control platform, and to monitor the communication quality data of the data link.

[0010] The storage layer is used to store the preset tunnel transportation path attribute data of the target tunnel, specifically the segment type labels pre-marked along different sections of the transportation route;

[0011] The decision-making layer includes a mode switching arbitration module and a driving mode control module, wherein the driving mode control module stores the control logic for unmanned driving mode and manned driving mode.

[0012] The mode switching arbitration module performs multi-source fusion positioning processing based on positioning observation data and motion state data to generate real-time positioning information of the dual-head transport vehicle in the tunnel; it performs environmental perception and recognition processing based on environmental observation data and motion state data to generate scene environment information containing information on passable areas and obstacle attributes; and it collaboratively calculates and generates a dynamic safety envelope containing the upper limit of safe vehicle speed and the upper limit of safe steering angle based on motion state data and scene environment information.

[0013] Based on the matching relationship between real-time positioning information and segment type labels, the current segment type and the segment type ahead are obtained. The system performs a comprehensive evaluation by coordinating scene environment information, dynamic safety envelope, and communication quality data. Based on this, a status evaluation matrix is ​​constructed to determine the target driving mode that matches the current and forward transportation conditions from the driving mode control module, and outputs control commands.

[0014] The execution layer is used to receive and execute control instructions.

[0015] Furthermore, the dual-head transport vehicle adopts a symmetrical dual-head body structure, with a driving control mechanism and a perception sensor group at the front and rear ends respectively. The perception sensor group specifically includes a lidar, millimeter-wave radar, vision camera, UWB ultra-wideband positioning module, IMU inertial measurement unit, and wheel speed sensor. The perception sensor group is used to collect perception data and motion status data.

[0016] Furthermore, the positioning observation data includes UWB positioning data obtained by the interaction between the UWB ultra-wideband positioning module and the preset UWB base station in the tunnel, laser point cloud data collected by lidar, and visual image data collected by the vision camera; the environmental observation data includes laser point cloud data collected by lidar, radar echo data collected by millimeter-wave radar, and visual image data collected by the vision camera; the motion state data includes acceleration and angular velocity data collected by the IMU inertial measurement unit, and the current vehicle speed collected by the wheel speed sensor.

[0017] Furthermore, when the mode switching arbitration module performs multi-source fusion positioning processing, it takes UWB positioning data, laser point cloud data, acceleration data, angular velocity data, current vehicle speed, and visual image data as inputs, and uses extended Kalman filtering or factor graph optimization algorithms to perform data fusion, outputting real-time positioning information of the dual-head transport vehicle in the tunnel without relying on satellite navigation signals.

[0018] Furthermore, when the mode switching arbitration module performs environmental perception and recognition processing, it performs time synchronization and spatial alignment of laser point cloud data, radar echo data, and visual image data.

[0019] The system performs target recognition on visual image data and outputs obstacle category labels and lane marking information as target recognition information. The obstacle category labels include category labels for static obstacles and dynamic obstacles.

[0020] Clustering and boundary extraction are performed on laser point cloud data and radar echo data. Combined with target recognition information, the system identifies the passable area and obstacles ahead, obtaining the left and right boundary positions of the passable area as passable area information. Simultaneously, the relative position and size of obstacles in the vehicle coordinate system are obtained as obstacle reference information. Based on the obstacle reference information, the system continuously tracks and obtains the speed and direction of motion of dynamic obstacles as obstacle motion information. The obstacle reference information and obstacle motion information are then bound to corresponding obstacle category labels to generate obstacle attribute information.

[0021] Based on angular velocity data and current vehicle speed, the global pose change of the dual-head transport vehicle at the moment of perception data acquisition is calculated as vehicle motion state information;

[0022] By utilizing the global pose change of the dual-headed transport vehicle, the relative position in the obstacle attribute information is transformed from the vehicle coordinate system to the global coordinate system for output. The transformed obstacle attribute information and passable area information are used together as scene environment information.

[0023] Furthermore, the mode switching arbitration module determines the vehicle's forward driving path based on the left and right boundary positions of the passable area in the relative position information, and calculates the width of the passable area.

[0024] Determine whether the width of the passable area is less than the sum of the width of the dual-headed transport vehicle and the preset safety margin. If so, mark the scene environment information as a path-restricted state.

[0025] Based on the obstacle's position and size information and obstacle category label in the obstacle attribute information, the following judgments are made: When the obstacle category label is a static obstacle, it is directly determined whether the current position of the static obstacle is on the vehicle's driving path. If so, the scene environment information is marked as an obstacle blocking state. When the obstacle category label is a dynamic obstacle, based on the obstacle's movement speed and direction in the obstacle movement information, combined with the current vehicle speed, the predicted position of the dynamic obstacle within a preset time window is calculated, and it is determined whether the predicted position intersects or overlaps with the vehicle's driving path. If so, the scene environment information is marked as an obstacle blocking state.

[0026] Furthermore, the dynamic security envelope is generated as follows:

[0027] The mode switching arbitration module reads the marked status and motion status data of the scene environment information;

[0028] When the scene environment information is marked as an obstacle blocking state, the current vehicle speed is obtained, and the safe braking distance at the current vehicle speed is calculated according to the vehicle dynamics model; the relative position of the obstacle in front in the vehicle coordinate system is obtained, the real-time obstacle distance at the current vehicle speed is calculated, and it is determined whether the safe braking distance is greater than the real-time obstacle distance. If so, the current vehicle speed is reduced until the safe braking distance is less than or equal to the real-time obstacle distance, and the reduced current vehicle speed is used as the upper limit of the safe vehicle speed.

[0029] When the scene environment information is marked as a path-restricted state, the maximum allowable steering angle required for the vehicle to pass through the passable area is calculated according to the vehicle kinematic model based on the width of the passable area, the wheelbase and track width of the dual-head transport vehicle, the preset inner wheel difference and outer contour sweep envelope constraints when the vehicle turns, and this maximum allowable steering angle is used as the upper limit of the safe steering angle.

[0030] Furthermore, the segment types include one or more of the following:

[0031] Long-distance straight transportation sections, tunnel boring machine assembly areas, segment alignment operation areas, areas rich in toxic and harmful gases, low visibility sections, and areas where multiple trades are working simultaneously.

[0032] The mode switching arbitration module performs a location query in the preset transportation path attribute data based on the real-time location information, takes the segment type corresponding to the location of the real-time location information as the current segment type, and takes the segment type that will be entered within a predetermined distance ahead of the real-time location information as the forward segment type.

[0033] Furthermore, the decision-making process for the target driving mode is as follows:

[0034] S1. Compare the communication quality data with the preset communication quality threshold to generate a communication status label. Compare the upper limit of the safe speed in the dynamic safety envelope with the current speed to generate a safety envelope status label. Determine the environment status label based on the marked status of the scene environment information. Determine the section status label based on the query result of the forward section type. The above four status labels constitute a status evaluation matrix.

[0035] S2. Based on the combination of state labels in the state evaluation matrix, select the autonomous driving mode or the manned driving mode as the target driving mode according to the pre-set priority order.

[0036] Furthermore, when the target driving mode is manned driving mode, the decision-maker activates the corresponding driving control mechanism based on the vehicle's current direction of travel; when the target driving mode is unmanned driving mode, the decision-maker locks both the front and rear driving control mechanisms of the vehicle.

[0037] Compared with the prior art, the advantages of this invention are:

[0038] 1. This invention constructs a multi-level control system architecture comprising a perception layer, a communication layer, a storage layer, a decision-making layer, and an execution layer. This architecture enables TBM tunnel transport vehicles to achieve high-precision positioning and full-condition environmental perception even in the absence of satellite navigation signals. Based on comprehensive physical perception, a progressive driving mode decision chain is constructed, consisting of "environmental perception marking → dynamic safety envelope calculation → priority decision of state evaluation matrix". This chain autonomously decides the target driving mode, enabling adaptive optimal switching between manned and unmanned driving modes based on real-time transportation conditions. This significantly improves the safety and intelligence level of tunnel transportation to a certain extent.

[0039] 2. This invention also deeply couples the dual-head vehicle body structure with the dual-mode control system, enabling efficient bidirectional driving without turning around in narrow tunnels and intelligent collaborative control between the two driver's seats:

[0040] Specifically, the dual-head transport vehicle adopts a completely symmetrical dual-head body structure, with independent driving control mechanisms and perception sensor groups at the front and rear ends. When the mode switching arbitration module selects the manned driving mode, the decision-making level automatically activates the corresponding driving control mechanism according to the vehicle's current direction of travel and performs an electrical signal lock on the driving control mechanism at the rear end of the direction of travel, ensuring that only the driver's seat at the front end of the direction of travel is effective and avoiding conflicts caused by simultaneous operation at both ends. When the unmanned driving mode is selected, the decision-making level locks both the driving control mechanisms at the front and rear ends of the vehicle to eliminate any risk of accidental human contact.

[0041] This collaborative control mechanism of "mode adjudication → corresponding end activation / locking" enables vehicles to achieve bidirectional driving without turning around in the narrow and confined space of a TBM tunnel, thus solving the technical problems of traditional single-head transport vehicles having difficulty turning around and the danger of reversing in tunnels, and greatly improving the mobility and safety of tunnel transportation. Attached Figure Description

[0042] Figure 1 This is a block diagram illustrating the module principle of the present invention;

[0043] Figure 2 This is a flowchart of the decision-making layer of the present invention;

[0044] Figure 3 This is a schematic diagram of a dual-head transport vehicle with a dual-head body structure according to the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0046] Example 1: This invention discloses a dual-mode control system for TBM tunnel electric dual-head transport vehicles, which supports both manned and unmanned driving modes. This system is applied to dual-head transport vehicles with a dual-head body structure. Please refer to [link / reference]. Figure 1 , Figure 2 and Figure 3 It includes a perception layer, a communication layer, a decision-making layer, and an execution layer. The decision-making layer includes a mode switching arbitration module and a driving mode control module, while the execution layer includes a steer-by-wire module, a brake-by-wire module, and a drive-by-wire module.

[0047] The dual-head transport vehicle adopts a symmetrical dual-head body structure. The front and rear ends are equipped with driving operation mechanisms and sensing sensor groups, and the structures at both ends are completely identical and arranged in a completely symmetrical manner. The driving operation mechanism at either end can independently complete the driving control of the vehicle.

[0048] The perception sensor group collects perception data of the target tunnel transportation scene and motion status data of the dual-head transport vehicle for the perception layer. The perception data includes positioning observation data and environmental observation data, which are uniformly sent to the mode switching arbitration module of the decision layer.

[0049] The perception sensor group specifically includes lidar, millimeter-wave radar, vision camera, UWB ultra-wideband positioning module, IMU inertial measurement unit and wheel speed sensor. The perception sensor group collects perception data of the target tunnel transportation scene and motion status data of the dual-head transport vehicle. The perception data includes positioning observation data and environmental observation data.

[0050] Specifically: the positioning observation data includes UWB positioning data obtained by the interaction between the UWB ultra-wideband positioning module and the preset UWB base station in the tunnel, laser point cloud data collected by lidar, and visual image data collected by visual camera;

[0051] Environmental observation data includes laser point cloud data collected by lidar, radar echo data collected by millimeter-wave radar, and visual image data collected by visual cameras;

[0052] Motion status data includes acceleration and angular velocity data collected by the IMU inertial measurement unit, and the current vehicle speed collected by the wheel speed sensor.

[0053] The communication layer is used to establish a data link between the dual-head transport vehicle and the ground control platform, and to monitor the communication quality data of the data link, specifically including communication latency (in milliseconds) and packet loss rate (percentage). The communication quality data is sent to the mode switching arbitration module of the decision layer in real time.

[0054] The storage layer stores the preset tunnel transportation path attribute data of the target tunnel. The tunnel transportation path attribute data is stored in the non-volatile memory of the storage layer in the form of an electronic map. Specifically, it consists of pre-marked section type labels along different sections of the transportation route. The section types include one or more of the following: long-distance straight transportation section (straight transportation section with a length greater than 500 meters, without curves, and without cross-operations), shield machine assembly area (located within 50 meters behind the shield machine, where segment assembly operations are required), segment alignment operation area (used for segment hoisting and alignment, requiring precise manual operation), toxic and harmful gas enrichment area (sections where methane or hydrogen sulfide gas may be enriched as indicated by geological surveys), low visibility section (sections with high dust concentration or insufficient lighting), and multi-trade cross-construction area (areas where multiple construction teams work simultaneously, with frequent personnel activity and complex game theory).

[0055] The decision-making layer includes a mode switching arbitration module and a driving mode control module, wherein the driving mode control module stores the control logic for unmanned driving mode and manned driving mode.

[0056] The mode switching arbitration module is used to receive positioning observation data, environmental observation data, motion status data, communication quality data, and segment type labels;

[0057] The mode switching arbitration module performs multi-source fusion positioning processing based on positioning observation data and motion state data to generate real-time positioning information of the dual-head transport vehicle in the tunnel; it performs environmental perception and recognition processing based on environmental observation data to generate scene environment information containing target recognition information, relative position information, and motion state information; and it collaboratively calculates and generates a dynamic safety envelope containing the upper limit of safe vehicle speed and the upper limit of safe steering angle based on motion state data and relative position information.

[0058] Based on the matching relationship between real-time positioning information and segment type labels, the current segment type and the segment type ahead are obtained. The system performs a comprehensive evaluation by coordinating scene environment information, dynamic safety envelope, and communication quality data. Based on this, a status evaluation matrix is ​​constructed to determine the target driving mode that matches the current and forward transportation conditions from the driving mode control module, and outputs control commands.

[0059] Among them, when the mode switching arbitration module performs environmental perception and recognition processing, it performs time synchronization and spatial alignment of laser point cloud data, radar echo data and visual image data.

[0060] The visual image data is used for target recognition, outputting obstacle category labels and lane marking information as target recognition data. The obstacle category labels are used to distinguish between static and dynamic obstacles. This process employs a deep learning-based target detection algorithm to process the visual images. The algorithm model has been pre-trained using an image dataset from a tunnel environment, containing no fewer than 10,000 labeled images covering categories such as personnel, transport vehicles, construction equipment, tunnel segment stacks, and loose objects. Target recognition outputs a category label for each detected target (including static obstacles: such as falling rocks, parked equipment, and tunnel segment stacks; dynamic obstacles: such as moving personnel, construction vehicles, and rail transport vehicles). Lane marking information includes the pixel position and type (solid line, dashed line, double line) of the lane lines, obtained through a semantic segmentation algorithm.

[0061] Clustering (using a density-based spatial clustering algorithm) and boundary extraction (using a random sampling consensus algorithm) are performed on laser point cloud data and radar echo data. Combined with target recognition information, the system identifies the passable area and obstacles ahead, obtaining the left and right boundary positions of the passable area as passable area information. Simultaneously, the relative position and size of obstacles in the vehicle coordinate system are obtained as obstacle reference information. Based on the obstacle reference information, the system continuously tracks and obtains the speed and direction of motion of dynamic obstacles as obstacle motion information. The obstacle reference information and obstacle motion information are then bound to corresponding obstacle category labels to generate obstacle attribute information.

[0062] Clustering is performed using a density-based spatial clustering algorithm. After clustering, each cluster represents an independent obstacle. The following attributes of the cluster are extracted: its relative position in the vehicle coordinate system (represented by the geometric center of the cluster, including lateral and longitudinal distances) and its size (length and width of the circumscribed rectangle of the cluster in the X and Y axes). Boundary extraction uses a random sampling consensus algorithm to perform planar fitting on the point cloud. The fitted planar boundaries are the left and right boundary positions of the passable area ahead, represented by lateral coordinate values ​​in the vehicle coordinate system. The identification of the passable area ahead and the obstacle ahead is also verified in conjunction with the target recognition information from the visual image.

[0063] Continuous tracking based on obstacle reference information employs a multi-target tracking algorithm (Kalman filtering or joint probabilistic data association). It performs association matching on the clustering results of the same obstacle identified in multiple consecutive frames (at least 10 frames). The matching criteria are the proximity (Euclidean distance less than 0.5 meters) and size similarity (size difference less than 20%) between obstacles. By calculating the removal of the same obstacle between adjacent frames at time intervals, the movement speed (in m / s) and movement direction (angle relative to the vehicle's direction of travel, in degrees) of the obstacle are obtained.

[0064] Based on angular velocity data and current vehicle speed, the global pose change of the dual-head transport vehicle at the moment of perception data acquisition is calculated as vehicle motion state information. Specifically, by performing time integration on the angular velocity data between two adjacent frames, the change in the heading angle of the vehicle within that time interval is obtained, which is the angle of rotation of the vehicle's head direction around the vertical axis.

[0065] At the same time, the arc length of the vehicle along its longitudinal axis is obtained by multiplying the current vehicle speed by the time interval. This arc length is then decomposed into the X-axis and Y-axis directions of the global coordinate system according to the change of the heading angle, so as to obtain the X-direction displacement and Y-direction displacement of the vehicle in the global coordinate system respectively.

[0066] Combining the changes in heading angle, displacement in the X direction, and displacement in the Y direction constitutes the complete pose change of the vehicle in the global coordinate system. This pose change describes the vehicle's position movement and attitude rotation in the global coordinate system (tunnel geodetic coordinate system) from the previous sampling time to the current sampling time.

[0067] By utilizing the global pose change of the dual-headed transport vehicle, the relative position in the obstacle attribute information is transformed from the vehicle coordinate system to the global coordinate system for output. Its core function is to give the obstacle a unique and fixed position description in the tunnel space, thereby realizing the spatiotemporal correlation of multi-frame data and the construction of global situation; the transformed obstacle attribute information and passable area information are used together as scene environment information.

[0068] The mode switching arbitration module determines the vehicle's forward driving path based on the left and right boundary positions of the passable area ahead, and calculates the width of the passable area; it determines whether the width of the passable area is less than the sum of the width of the dual-head transport vehicle and the preset safety margin. If so, it marks the scene environment information as a path-restricted state.

[0069] Based on the obstacle's position and size information and obstacle category label in the obstacle attribute information, the following judgments are made: When the obstacle category label is a static obstacle, it is directly determined whether the current position of the static obstacle is on the vehicle's driving path. If so, the scene environment information is marked as an obstacle blocking state. When the obstacle category label is a dynamic obstacle, based on the obstacle's movement speed and direction in the obstacle movement information, combined with the current vehicle speed, the predicted position of the dynamic obstacle within a preset time window is calculated. It is then determined whether the predicted position intersects or overlaps with the vehicle's driving path. If so, the scene environment information is marked as an obstacle blocking state.

[0070] The method for calculating the predicted position of dynamic obstacles is as follows: Assuming that the dynamic obstacle maintains uniform linear motion within a preset time window, and the duration of the preset time window is dynamically adjusted according to the current vehicle speed, then the predicted position = the relative position of the obstacle in front + the speed of movement × the duration of the time window × the preset unit vector of the direction of movement. The unit vector of the direction of movement refers to a direction vector with a length of 1 that describes the forward direction of the dynamic obstacle. The predicted position is then converted to the vehicle coordinate system.

[0071] If the absolute value of the lateral offset of the predicted position is less than half the width of the vehicle and half the width of the obstacle, and the longitudinal distance of the predicted position is less than the preset detection distance threshold, then it is determined that the predicted position of the dynamic obstacle intersects or overlaps with the driving path in front of the vehicle. Here, the lateral offset of the predicted position refers to the vertical distance from the predicted position of the obstacle to the longitudinal centerline of the vehicle (i.e., the axis of the vehicle's direction of travel), and the detection distance threshold refers to the upper limit of the effective detection range along the longitudinal direction of the vehicle.

[0072] The dynamic security envelope is generated as follows:

[0073] The mode switching arbitration module reads the marked status and motion status data of the scene environment information;

[0074] When the scene environment information is marked as an obstacle blocking state, the current vehicle speed is obtained, and the safe braking distance at the current vehicle speed is calculated according to the vehicle dynamics model; the relative position of the obstacle in front in the vehicle coordinate system is obtained, and the real-time obstacle distance at the current vehicle speed is calculated. The real-time obstacle distance = the longitudinal distance of the obstacle - the distance offset from the current position of the vehicle to the front of the vehicle. This distance offset is pre-calibrated by the structural dimensions of the vehicle (e.g., 1.5 meters) and is a fixed constant. The longitudinal distance of the obstacle is obtained from the position information in the obstacle attribute information.

[0075] Determine if the safe braking distance is greater than the real-time obstacle distance. If so, reduce the current vehicle speed until the safe braking distance is less than or equal to the real-time obstacle distance. Use the reduced current vehicle speed as the safe speed limit. The physical meaning of the safe speed limit is: under the current road conditions, the highest permissible vehicle speed that can guarantee safe braking and stopping in front of the obstacle.

[0076] When the scene environment information is marked as a path-restricted state, a vehicle kinematic model is established based on the Ackermann steering geometry principle according to the width of the passable area and the wheelbase and track width of the dual-head transport vehicle. Combined with the preset inner wheel difference and outer contour sweep envelope constraints when the vehicle turns, the maximum allowable steering angle required for the vehicle to pass through the passable area is calculated according to the vehicle kinematic model, which serves as the upper limit of the safe steering angle.

[0077] Among them, the inner wheel difference is characterized by the radius difference between the inner rear wheel trajectory and the outer front wheel trajectory, the outer contour sweep envelope is characterized by the maximum lateral span of the turning trajectory of the four corner points of the vehicle's rectangular boundary (length = wheelbase, width = track width) at a given steering angle, and the physical meaning of the upper limit of the safe steering angle is the maximum steering angle allowed under the current channel width. Exceeding this angle will cause the outer front wheel or inner rear wheel of the vehicle to go beyond the boundary of the passable area, which may result in the risk of scraping the tunnel wall or colliding.

[0078] The mode switching arbitration module performs a location query in the preset transportation path attribute data based on the real-time location information, takes the segment type corresponding to the location of the real-time location information as the current segment type, and takes the segment type that will be entered within a predetermined distance ahead of the real-time location information as the forward segment type.

[0079] The determination process for the target driving mode is as follows:

[0080] S1. Compare the communication quality data with the preset communication quality threshold to generate a communication status label. Compare the upper limit of the safe speed in the dynamic safety envelope with the current speed to generate a safety envelope status label. Determine the environment status label based on the marked status of the scene environment information. Determine the section status label based on the query result of the forward section type. The above four status labels constitute a status evaluation matrix.

[0081] The specific logic for generating communication status tags is as follows: when the communication latency is less than the first preset threshold of 50ms and the packet loss rate is less than the second preset threshold of 5%, it is marked as a normal communication state; otherwise, it is marked as an abnormal communication state.

[0082] The specific generation logic of the safety envelope status label is as follows: compare the upper limit of safe speed (in km / h) in the dynamic safety envelope with the current speed (in km / h). When the upper limit of safe speed is greater than or equal to the current speed, it is marked as a safety envelope satisfied state; otherwise, it is marked as a safety envelope exceeded state.

[0083] The specific generation logic of the environment status label is as follows: it is determined based on the labeling status of the scene environment information. When it is not labeled as a path-restricted state and not labeled as an obstacle-blocking state, it is labeled as an environment passable state; when it is labeled as a path-restricted state, it is labeled as a path-restricted state; when it is labeled as an obstacle-blocking state, it is labeled as an obstacle-blocking state.

[0084] The specific logic for generating section status labels is as follows: it is determined based on the query results of the preceding section type. When the preceding section type is a long-distance straight-line transportation section, it is marked as a regular section; when the preceding section type is a tunnel boring machine assembly area, segment alignment operation area, or multi-trade cross-construction area, it is marked as a complex operation section; when the preceding section type is an area rich in toxic and harmful gases or a low-visibility section, it is marked as a hazardous environment section.

[0085] S2. Based on the combination of state labels in the state evaluation matrix, select the unmanned driving mode or the manned driving mode as the target driving mode according to the pre-set priority order.

[0086] The pre-set priority order is as follows:

[0087] Emergency safety priority: When the environmental status label is obstructed by obstacles, the safety envelope status label is exceeded, or the communication status label is abnormal, select the manned driving mode as the target driving mode.

[0088] Prioritize complex working conditions: When the environmental status label is path-restricted or the section status label is complex work section, select the manned driving mode as the target driving mode.

[0089] Conventional autonomous driving: When the communication status label is normal, the safety envelope status label is satisfied, the environment status label is passable, and the section status label is a conventional section, the autonomous driving mode is selected as the target driving mode.

[0090] Autonomous driving in hazardous environments: When the segment status label is a hazardous environment segment, the communication status label is a normal communication status, and the safety envelope status label is a safety envelope satisfied status, the autonomous driving mode is selected as the target driving mode.

[0091] Example 2: The execution layer receives and executes control commands for the corresponding target driving mode. Simultaneously, the arbitration result of the mode switching module directly drives the physical actions of the dual-head body structure.

[0092] When the target driving mode is manned driving mode, the decision-maker activates the corresponding driving control mechanism according to the vehicle's current direction of travel; when the target driving mode is unmanned driving mode, the decision-maker locks the driving control mechanisms at both the front and rear of the vehicle.

[0093] This mechanism ensures that:

[0094] Manned driving mode: Only the driver's seat at the front of the direction of travel is effective, while the driver's seat at the rear is locked by an electrical signal to prevent conflict caused by simultaneous operation from both ends;

[0095] Autonomous driving mode: All driving control mechanisms at both ends are locked to eliminate any risk of accidental human intervention;

[0096] In autonomous driving mode, the steering angle command issued by the execution layer shall not exceed the upper limit of the safe steering angle.

[0097] In addition, the steer-by-wire module, brake-by-wire module and drive-by-wire module in the execution layer each perform internal channel self-checks in real time, and use the available steering torque range, available braking force range and available drive power range of each actuator obtained from the self-check as actuator availability performance constraints to feed back to the mode switching arbitration module of the decision layer.

[0098] The mode switching arbitration module corrects the dynamic safety envelope based on the actuator availability performance constraints: it recalculates the maximum safe braking deceleration based on the available braking force range and updates the upper limit of safe vehicle speed with the recalculated maximum safe braking deceleration; it also recalculates the maximum safe steering angle based on the available steering torque range and updates the upper limit of safe steering angle with the recalculated maximum safe steering angle.

[0099] The various thresholds and preset values ​​involved in this invention (including but not limited to communication delay threshold, packet loss rate threshold, vehicle speed safety envelope comparison value, safety margin of passable area width and vehicle width, obstacle detection distance threshold, dynamic obstacle prediction time window, speed limit threshold under path-restricted conditions, etc.) are all conventional design choices or engineering experience values ​​taken by those skilled in the art. Their specific values ​​can be determined through a limited number of tests or experience based on different tunnel construction conditions, vehicle parameters (such as full load weight, wheelbase, track width, braking performance) and safety standards, and do not constitute a limitation on the scope of protection of this invention.

[0100] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto; any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and its improved concept, should be covered within the scope of protection of the present invention.

Claims

1. A dual-mode control system for manned and unmanned driving of a TBM (Tunnel Boring Machine) electric dual-head transport vehicle, applied to dual-head transport vehicles with a dual-head body structure, characterized by: include: The perception layer is used to collect perception data of the target tunnel transportation scenario and motion status data of the dual-head transport vehicle. The perception data includes positioning observation data and environmental observation data. The communication layer is used to establish a data link between the dual-head transport vehicle and the ground control platform, and to monitor the communication quality data of the data link. The storage layer is used to store the preset tunnel transportation path attribute data of the target tunnel, specifically the segment type labels pre-marked along different sections of the transportation route; The decision-making layer includes a mode switching arbitration module and a driving mode control module, wherein the driving mode control module stores the control logic for unmanned driving mode and manned driving mode. The mode switching arbitration module performs multi-source fusion positioning processing based on positioning observation data and motion state data to generate real-time positioning information of the dual-head transport vehicle in the tunnel; performs environmental perception and recognition processing based on environmental observation data and motion state data to generate scene environment information including passable area information and obstacle attribute information; and collaboratively calculates and generates a dynamic safety envelope including the upper limit of safe vehicle speed and the upper limit of safe steering angle based on motion state data and scene environment information. Based on the matching relationship between real-time positioning information and segment type labels, the current segment type and the segment type ahead are obtained. The system performs a comprehensive evaluation by coordinating scene environment information, dynamic safety envelope, and communication quality data. Based on this, a state evaluation matrix is ​​constructed. The system then decides whether to use the unmanned driving mode or the manned driving mode as the target driving mode from the driving mode control module and outputs control commands. The execution layer is used to receive and execute control instructions.

2. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 1, characterized in that: The dual-head transport vehicle adopts a symmetrical dual-head body structure, with a driving control mechanism and a perception sensor group at the front and rear ends respectively. The perception sensor group specifically includes a lidar, millimeter-wave radar, vision camera, UWB ultra-wideband positioning module, IMU inertial measurement unit and wheel speed sensor. The perception sensor group is used to collect perception data and motion status data.

3. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 1, characterized in that: The positioning observation data includes UWB positioning data, laser point cloud data, and visual image data; the environmental observation data includes laser point cloud data, radar echo data, and visual image data; the motion state data includes acceleration data, angular velocity data, and current vehicle speed.

4. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 1, characterized in that: When the mode switching arbitration module performs multi-source fusion positioning processing, it uses the positioning observation data as input and employs extended Kalman filtering or factor graph optimization algorithms to perform data fusion, outputting the real-time positioning information of the dual-head transport vehicle in the tunnel.

5. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 4, characterized in that: When the mode switching arbitration module performs environmental perception and recognition processing, it synchronizes and aligns the environmental observation data in time and space; it performs target recognition on the visual image data and outputs obstacle category labels and lane marking information as target recognition information. The obstacle category labels include static obstacles and dynamic obstacles. Clustering and boundary extraction are performed on laser point cloud data and radar echo data. Collaborative target recognition information is used to identify passable areas and obstacles ahead, and the left and right boundary positions of the passable areas ahead are obtained as passable area information. At the same time, the relative position and size of the obstacles ahead in the vehicle coordinate system are obtained as obstacle reference information. Based on the obstacle reference information, the movement speed and direction of the dynamic obstacles are continuously tracked and obtained as obstacle movement information. By binding obstacle baseline information and obstacle motion information with corresponding obstacle category labels, obstacle attribute information is generated; Based on the angular velocity data and the current vehicle speed, the global pose change of the dual-head transport vehicle at the moment of perception data acquisition is calculated. Based on the global pose change, the relative position in the obstacle attribute information is transformed from the vehicle coordinate system to the global coordinate system and output. The transformed obstacle attribute information and passable area information are used together as scene environment information.

6. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 5, characterized in that: The mode switching arbitration module determines the vehicle's forward driving path based on the left and right boundary positions of the passable area ahead, and calculates the width of the passable area, thereby determining whether to mark the scene environment information as a path-restricted state. Based on the obstacle's location and size information and obstacle category label in the obstacle attribute information, determine whether to mark the scene environment information as an obstacle blocking state.

7. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 6, characterized in that: The dynamic security envelope is generated as follows: The mode switching arbitration module reads the marked status and motion status data of the scene environment information; When the scene environment information is marked as an obstacle blocking state, the current vehicle speed is obtained, and the safe braking distance at the current vehicle speed is calculated according to the vehicle dynamics model; the relative position of the obstacle in front in the vehicle coordinate system is obtained, the real-time obstacle distance at the current vehicle speed is calculated, and it is determined whether the safe braking distance is greater than the real-time obstacle distance. If so, the current vehicle speed is reduced until the safe braking distance is less than or equal to the real-time obstacle distance, and the reduced current vehicle speed is taken as the upper limit of the safe vehicle speed. When the scene environment information is marked as a path-restricted state, the maximum permissible steering angle required for the vehicle to pass through the passable area is calculated according to the vehicle kinematic model based on the width of the passable area, the wheelbase and track width of the dual-head transport vehicle, the preset inner wheel difference and outer contour sweep envelope constraints when the vehicle turns, and this maximum permissible steering angle is used as the upper limit of the safe steering angle.

8. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 7, characterized in that: The segment type includes one or more of the following: Long-distance straight transportation section, tunnel boring machine assembly area, segment alignment operation area, toxic and harmful gas enrichment area, low visibility section, multi-trade cross construction area; The mode switching arbitration module performs a location query in the preset transportation path attribute data based on the real-time location information, takes the segment type corresponding to the location of the real-time location information as the current segment type, and takes the segment type that will be entered within a predetermined distance ahead of the real-time location information as the forward segment type.

9. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 8, characterized in that: The determination process for the target driving mode is as follows: S1. Compare the communication quality data with the preset communication quality threshold to generate a communication status label. Compare the upper limit of the safe speed in the dynamic safety envelope with the current speed to generate a safety envelope status label. Determine the environment status label based on the marked status of the scene environment information. Determine the section status label based on the query result of the forward section type. The above four status labels constitute a status evaluation matrix. S2. Based on the combination of state labels in the state evaluation matrix, select the autonomous driving mode or the manned driving mode as the target driving mode according to the pre-set priority order.

10. The TBM tunnel electric dual-head transport vehicle manned and unmanned dual-mode control system according to claim 9, characterized in that: When the target driving mode is manned driving mode, the decision-maker activates the corresponding driving control mechanism according to the vehicle's current direction of travel; when the target driving mode is unmanned driving mode, the decision-maker locks the driving control mechanisms at both the front and rear of the vehicle.

Images