Land-air dual-purpose tunnel inspection robot intelligent control system

Abstract

The invention relates to an air-ground dual-purpose tunnel inspection robot intelligent control system, and belongs to the technical field of robot control, the system comprises an unmanned aerial vehicle and an unmanned vehicle, the unmanned aerial vehicle and the unmanned vehicle are cooperatively used, and cooperative use of the unmanned aerial vehicle and the unmanned vehicle comprises control of integrated operation of the unmanned aerial vehicle and the unmanned vehicle and control of split operation of the unmanned aerial vehicle and the unmanned vehicle; the system further comprises a sensing fusion module, a path planning module, a network communication module, an obstacle observation and prediction module and an intelligent control module, the sensing fusion module is arranged on the unmanned aerial vehicle and the unmanned vehicle, and the sensing fusion module is used for detecting external information of the unmanned aerial vehicle and the unmanned vehicle. The sensing fusion module comprises a camera, a radar, a laser radar, an ultrasonic sensor, an inertial measurement sensor, a GPS positioning sensor, a smoke alarm sensor and a toxic gas detection sensor; the unmanned aerial vehicle communicates with the unmanned vehicle; according to the invention, the safety and reliability of tunnel inspection work are effectively improved, and accurate control of the tunnel inspection robot is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of robot control, and in particular relates to an intelligent control system for a land and air dual-purpose tunnel inspection robot. Background Art

[0002] With the acceleration of urbanization, underground tunnel cables and other facilities are important components of urban infrastructure, and their safe and stable operation is crucial to ensure the normal functioning of urban functions. However, the internal environment of the tunnel is complex and changeable. It is not only limited by natural conditions such as small physical space, dim light, and poor air circulation, but also faces potential safety hazards such as cable aging and structural damage. Therefore, tunnel inspection is very necessary. The traditional manual tunnel inspection method is inefficient, and the safety risks of operators in harsh environments are high. It is difficult to meet the efficient and safe requirements of modern cities for tunnel inspection and maintenance.

[0003] In recent years, with the development of robot technology, tunnel inspection robots have begun to be used in daily inspection and maintenance work in underground spaces to improve inspection efficiency, safety and reduce labor costs. Despite this, existing inspection robots still face many challenges in practical application. First, the interior of the tunnel is usually in a state of satellite signal denial, which causes the navigation system based on satellite positioning to fail, affecting the robot's autonomous path navigation capability; second, the existing inspection robots are mainly ground mobile inspections, and the complex terrain conditions limit the robot's mobility, especially when encountering obstacles or narrow passages; finally, the detection conditions in the tunnel are harsh, requiring the robot to have highly sensitive perception and adaptability. Due to the complex and changeable environment in the tunnel, it is not conducive to the inspection work of the inspection robot, which brings difficulties to the precise control of the inspection robot during the inspection process.

[0004] To sum up, it is necessary to develop an intelligent control system for a land-air dual-purpose tunnel inspection robot that can overcome the problems of satellite signal denial, limited mobility, and harsh detection conditions in the above-mentioned tunnels, so as to achieve precise control of the tunnel inspection robot. Summary of the invention

[0005] The present invention aims at the above-mentioned problems, makes up for the deficiencies of the prior art, and provides an intelligent control system for a land-air dual-purpose tunnel inspection robot; the intelligent control system for a land-air dual-purpose tunnel inspection robot of the present invention can effectively improve the safety and reliability of tunnel inspection work, and achieve the effect of precise control of the tunnel inspection robot.

[0006] To achieve the above purpose, the present invention adopts the following technical solution.

[0007] The present invention provides an intelligent control system for a dual-purpose tunnel inspection robot for land and air, wherein the dual-purpose tunnel inspection robot for land and air comprises a drone and an unmanned vehicle, which are used in coordination with the unmanned vehicle. The coordinated use of the drone and the unmanned vehicle comprises controlling the drone and the unmanned vehicle to operate in an integrated manner, and controlling the drone and the unmanned vehicle to operate in a separate manner. Controlling the drone and the unmanned vehicle to operate in an integrated manner: when the unmanned vehicle encounters a road section that is difficult to drive, the drone is controlled to land on the unmanned vehicle and be fixed between the unmanned vehicle, and the unmanned vehicle is driven by the drone to fly over the road section. Controlling the drone and the unmanned vehicle to operate in a separate manner: when on a normal road section, the drone and the unmanned vehicle are controlled to drive independently and conduct joint inspections. The intelligent control system includes a perception fusion module, a path planning module, a network communication module, an obstacle observation and prediction module, and an intelligent control module. The perception fusion module is respectively arranged on the UAV and the unmanned vehicle. The perception fusion module is used to detect the external state information of the UAV and the unmanned vehicle. The perception fusion module includes a camera, a radar, a laser radar, an ultrasonic sensor, an inertial measurement sensor, a GPS positioning sensor, a smoke alarm sensor, and a toxic gas detection sensor; the network communication module is used for mutual communication between the UAV and the unmanned vehicle; On normal roads, the path planning module plans the driving paths of the unmanned vehicle and the unmanned aerial vehicle respectively; at the same time, the obstacle observation and prediction module observes and predicts the position and status information of external obstacles, and feeds back the observed and predicted position and status information of external obstacles to the path planning module, so as to cooperate with the path planning module to plan the optimal driving paths for the unmanned vehicle and the unmanned aerial vehicle respectively; When the unmanned vehicle encounters a difficult-to-drive section of road, the path planning module plans the optimal driving path for the drone to land on the unmanned vehicle. After the drone lands on the unmanned vehicle, the drone and the unmanned vehicle are fixed by magnetic attraction to complete the control of the integrated operation of the drone and the unmanned vehicle; according to the optimal driving paths for the unmanned vehicle, drone, and the integration of unmanned vehicle and drone provided by the path planning module, the intelligent control module controls the unmanned vehicle, drone, and the unmanned vehicle and drone when they need to operate in an integrated manner to track the optimal driving paths.

[0008] As a preferred solution of the present invention, the operation of the integration process of the UAV and the unmanned vehicle is controlled by an integrated operation control model, and the expression of the integrated operation control model is as follows: ; Among them, for the UAV operation control model z i , u i It is expressed as: , ; For the unmanned vehicle operation control model z i , ui It is expressed as: , ;in, A , B , C is the coefficient matrix; X is the horizontal axis, Y is the vertical axis; T is the transposition symbol.

[0009] As another preferred embodiment of the present invention, the path planning module includes an A-star algorithm, and the path planning module uses the A-star algorithm to cooperate with the obstacle observation and prediction module to plan the optimal driving path for the unmanned vehicle and the unmanned aerial vehicle respectively, and the A-star algorithm is used for global planning; the obstacle observation and prediction module includes a tangent network model, and the obstacle observation and prediction module uses the tangent network model to observe and predict the position and state information of external obstacles. The input of the tangent network model is the external state information detected by the camera, radar, lidar, ultrasonic sensor, inertial measurement sensor, GPS positioning sensor, smoke alarm sensor, and toxic gas detection sensor. The output of the tangent network model is the obstacle position and state information, and the obstacle observation and prediction module feeds back the obstacle position and state information to the A-star algorithm, and the obstacle position and state information is used as the constraint condition of the A-star algorithm.

[0010] As another preferred embodiment of the present invention, the path planning module also includes a tangent network model; when the unmanned vehicle encounters a difficult-to-drive section of road, the path planning module plans the optimal driving path for the drone to land on the unmanned vehicle. After the drone lands on the unmanned vehicle, the drone and the unmanned vehicle are fixed by magnetic attraction to complete the integrated operation of the drone and the unmanned vehicle, specifically including: the unmanned vehicle transmits its own position information to the drone in real time, the path planning module uses the tangent network model to predict the driving trajectory of the unmanned vehicle, the input of the tangent network model is the historical position information of the unmanned vehicle, and the output is the predicted position of the unmanned vehicle, the drone path planning terminal position is updated in real time, and the path for the drone to land on the unmanned vehicle is planned in real time in combination with the A-star algorithm.

[0011] As another preferred embodiment of the present invention, the intelligent control system also includes a charging module, which is installed on the unmanned vehicle. When the drone needs to be charged, the path planning module plans the optimal driving path for the drone to land on the unmanned vehicle to complete the integrated operation of the drone and the unmanned vehicle. After the drone and the unmanned vehicle are integrated and fixed, the charging module charges the drone.

[0012] As another preferred embodiment of the present invention, the tangent network model is in the following form: ; in, is the tangent network model weight, , x is the network input; the above formula is written as t The general form of the dynamics equation of a dimensional system is: ; in, is the tangent network model weight, x For network input, l To expand the power, q is the number of items and , e is the approximation error, and T is the transposed sign.

[0013] As another preferred embodiment of the present invention, the tangent network model optimizes its parameters using a learning algorithm, and the learning algorithm adopts any one of a recursive least squares algorithm, a back propagation algorithm, and a wake-sleep algorithm.

[0014] As another preferred solution of the present invention, the intelligent control module adopts a tangent network controller, and the form of the tangent network controller is as follows: ; Among them, the controller u 1 Input is ; ; r 1 The optimal driving path for the integrated operation of drones and unmanned vehicles is provided by the path planning module; y 1 Output trajectory for the integration of drones and unmanned vehicles; is the weight of the tangent network model; x Input for the network; l To expand the power; q is the number of items, ; T is the transposition symbol; ; Among them, the controller u 2 Input is ; ; r 2 The optimal driving path for the unmanned vehicle is provided by the path planning module; y 2 Output trajectory for the unmanned vehicle; is the weight of the tangent network model; x Input for the network; l To expand the power; q is the number of items, ; T is the transposition symbol; ; Among them, the controller u 3 Input is ; ; r 3 The optimal driving path for the UAV is provided by the path planning module; y 3 Output trajectory for the drone; is the weight of the tangent network model; x Input for the network; l To expand the power; q is the number of items, ; T is the transposition symbol.

[0015] As another preferred solution of the present invention, the split operation control model is used to control the split operation of the drone and the unmanned vehicle, and the split operation control model includes an independent operation control model of the unmanned vehicle and an independent operation control model of the drone; The specific control model of the unmanned vehicle independent operation is as follows: ; in, x , y for location; v x , v y They are x axis, y Speed ​​in the axis direction; i is the yaw angle; oh is the angular velocity of rotation; The specific control model of independent operation of UAV is as follows: ; in, x , y , z is the spatial position; f , i , ψ They represent the roll angle, pitch angle and yaw angle respectively; I xx , I yy , I zz is the moment of inertia; m For drone quality; L is the distance between the motor shaft and the center of the fuselage; g represents the acceleration due to gravity; u 4 ,u 5 , u 6 , u 7 Input for drone control.

[0016] Beneficial effects of the present invention: The intelligent control system of a land and air dual-purpose tunnel inspection robot provided by the present invention can effectively improve the safety and reliability of tunnel inspection work and achieve the effect of precise control of the tunnel inspection robot through the combination of the integrated operation control model, the split operation control model, the perception fusion module, the path planning module, the network communication module, the obstacle observation and prediction module and the intelligent control module. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The present invention is a schematic structural diagram of an intelligent control system for a land-air dual-purpose tunnel inspection robot.

[0018] Markings in the figure: 1 is the perception fusion module, 101 is the camera, 102 is the radar, 103 is the lidar, 104 is the ultrasonic sensor, 105 is the inertial measurement sensor, 106 is the GPS positioning sensor, 107 is the smoke alarm sensor, and 108 is the toxic gas detection sensor; 2 is the integrated operation control model; 3 is the split operation control model; 4 is the path planning module; 5 is the intelligent control module; 6 is the network communication module; 7 is the obstacle observation and prediction module; 8 is the charging module; 9 is the land and air dual-purpose tunnel inspection robot, 901 is the drone, and 902 is the unmanned vehicle. DETAILED DESCRIPTION

[0019] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be understood that the specific implementation methods described herein are only used to explain the present invention and are not used to limit the present invention.

[0020] Combination Figure 1As shown, an intelligent control system of a dual-purpose land and air tunnel inspection robot provided by an embodiment of the present invention, the dual-purpose land and air tunnel inspection robot 9 includes a drone 901 and an unmanned vehicle 902, the drone 901 and the unmanned vehicle 902 are used in conjunction with each other, and the coordinated use of the drone 901 and the unmanned vehicle 902 includes controlling the drone 901 and the unmanned vehicle 902 to operate in an integrated manner, and controlling the drone 901 and the unmanned vehicle 902 to operate in a separate manner; controlling the drone 901 and the unmanned vehicle 902 to operate in an integrated manner: when the unmanned vehicle 902 encounters a difficult-to-drive section of road, the drone 901 is controlled to land on the unmanned vehicle 902 and be fixed between the unmanned vehicle 902, and the unmanned vehicle is driven by the drone 901 902 flies over this road section; controls the UAV 901 and the unmanned vehicle 902 to operate separately: on normal roads, controls the UAV 901 and the unmanned vehicle 902 to drive independently and inspect together; the intelligent control system includes a perception fusion module 1, a path planning module 4, a network communication module 6, an obstacle observation and prediction module 7, and an intelligent control module 5. The perception fusion module 1 is respectively arranged on the UAV 901 and the unmanned vehicle 902. The perception fusion module 1 is used to detect the external state information of the UAV 901 and the unmanned vehicle 902. The perception fusion module 1 includes a camera 101, a radar 102, a laser radar 103, an ultrasonic sensor 104, and an inertial measurement sensor 105, GPS positioning sensor 106, smoke alarm sensor 107, toxic gas detection sensor 108; network communication module 6, used for communication between drone 901 and unmanned vehicle 902; in normal road sections, the path planning module 4 plans the driving path of unmanned vehicle 902 and the driving path of drone 901 respectively; at the same time, the obstacle observation and prediction module 7 observes and predicts the position and state information of external obstacles, and feeds back the observed and predicted position and state information of external obstacles to the path planning module 4, so as to cooperate with the path planning module 4 to realize the planning of the optimal driving path for unmanned vehicle 902 and drone 901 respectively; when there is no When the human vehicle encounters a difficult-to-drive section, the path planning module 4 plans the optimal driving path for the drone 901 to land on the unmanned vehicle 902. After the drone 901 lands on the unmanned vehicle 902, the drone 901 and the unmanned vehicle 902 are fixed by magnetic attraction to complete the control of the integrated operation of the drone 901 and the unmanned vehicle 902; according to the optimal driving path of the unmanned vehicle 902, the drone 901, and the integration of the unmanned vehicle 902 and the drone 901 provided by the path planning module, the intelligent control module 5 controls the unmanned vehicle 902, the drone 901, and the unmanned vehicle 902 and the drone 901 when they need to be operated in an integrated manner to track the optimal driving path.

[0021] Specifically, the operation of the integration process of the UAV 901 and the unmanned vehicle 902 is controlled by the integrated operation control model 2. The expression of the integrated operation control model is as follows: ; Among them, for the UAV 901 operation control model z i , u i It is expressed as: , ; For the unmanned vehicle 902 operation control model z i , u i It is expressed as: , ;in, A , B , C is the coefficient matrix; X is the horizontal axis, Y is the vertical axis; T is the transposition symbol.

[0022] Specifically, the path planning module 4 includes an A-star algorithm, which is also called the "A-star algorithm" or the "A-star algorithm". The A-star algorithm is a powerful and flexible heuristic search algorithm, which is particularly suitable for solving problems that require efficient finding of the optimal path; the path planning module 4 uses the A-star algorithm in conjunction with the obstacle observation and prediction module 7 to plan the optimal driving paths for the unmanned vehicle 902 and the unmanned aerial vehicle 901 respectively, and the A-star algorithm is used for global planning; the obstacle observation and prediction module 7 includes a tangent network model, and the obstacle observation and prediction module 7 uses a tangent network model to calculate the optimal driving paths for the unmanned vehicle 902 and the unmanned aerial vehicle 901. The network model observes and predicts the position and state information of external obstacles. The input of the tangent network model is the external state information detected by the camera 101, radar 102, lidar 103, ultrasonic sensor 104, inertial measurement sensor 105, GPS positioning sensor 106, smoke alarm sensor 107, and toxic gas detection sensor 108. The output of the tangent network model is the obstacle position and state information. The obstacle observation and prediction module 7 feeds back the obstacle position and state information to the A-star algorithm, and the obstacle position and state information is used as the constraint condition of the A-star algorithm.

[0023] Specifically, the path planning module 4 also includes a tangent network model; when the unmanned vehicle encounters a difficult-to-drive section of road, the path planning module 4 plans the optimal driving path for the drone to land on the unmanned vehicle. After the drone 901 lands on the unmanned vehicle 902, the drone 901 and the unmanned vehicle 902 are fixed by magnetic attraction to complete the control of the integrated operation of the drone 901 and the unmanned vehicle 902, specifically including: the unmanned vehicle 902 transmits its own position information to the drone 901 in real time, and the path planning module 4 uses the tangent network model to predict the driving trajectory of the unmanned vehicle. The input of the tangent network model is the historical position information of the unmanned vehicle 902, and the output is the predicted position of the unmanned vehicle 902. The end point position of the drone 901 path planning is updated in real time, and the path of the drone 901 landing on the unmanned vehicle 902 is planned in real time in combination with the A-star algorithm.

[0024] Specifically, the intelligent control system also includes a charging module 8, which is installed on the unmanned vehicle 902. When the drone 901 needs to be charged, the path planning module 4 is used to plan the optimal driving path for the drone 901 to land on the unmanned vehicle 902, so as to complete the control of the integrated operation of the drone 901 and the unmanned vehicle 902. After the drone 901 and the unmanned vehicle 902 are integrated and fixed, the charging module 8 charges the drone 901.

[0025] Specifically, the tangent network model takes the following form: ; in, is the tangent network model weight, , x is the network input; the above formula is written as t The general form of the dynamics equation of a dimensional system is: ; in, is the tangent network model weight, x For network input, l To expand the power, q is the number of items and , e is the approximation error, and T is the transposed sign.

[0026] Specifically, the tangent network model optimizes its parameters using a learning algorithm, and the learning algorithm uses any one of a recursive least squares algorithm, a back propagation algorithm, and a wake-sleep algorithm; the learning algorithm can also use other algorithms that can meet the optimization function, not limited to the above learning algorithms.

[0027] Specifically, the intelligent control module 5 adopts a tangent network controller, and the form of the tangent network controller is as follows: ; Among them, the controller u 1 Input is ; ; r 1 The optimal driving path for the integrated operation of the UAV 901 and the unmanned vehicle 902 is provided by the path planning module 4; y 1 Outputting trajectories for the integration of the UAV 901 and the unmanned vehicle 902; is the weight of the tangent network model; x Input for the network; l To expand the power; q is the number of items, ; T is the transposition symbol; ;

[0028] Among them, the controller u 2 Input is ; ; r 2 The optimal driving path of the unmanned vehicle 902 is provided by the path planning module 4; y 2 Output trajectory for unmanned vehicle 902; is the weight of the tangent network model; x Input for the network; l To expand the power; q is the number of items, ; T is the transposition symbol; ;

[0029] Among them, the controller u 3 Input is ; ; r 3 The optimal driving path of the UAV 901 is provided by the path planning module 4; y 3 Output trajectory for drone 901; is the weight of the tangent network model; x Input for the network; l To expand the power; q is the number of items, ; T is the transposition symbol.

[0030] Specifically, the split operation control model 3 is used to control the split operation of the drone 901 and the unmanned vehicle 902, and the split operation control model includes an independent operation control model of the unmanned vehicle 902 and an independent operation control model of the drone 901; The independent operation control model of unmanned vehicle 902 is as follows: ; in, x , y for location; v x , v y They are x axis, y Speed ​​in the axis direction; i is the yaw angle; oh is the angular velocity of rotation; The independent operation control model of UAV 901 is as follows: ; in, x , y , z is the spatial position; f , i , ψ They represent the roll angle, pitch angle and yaw angle respectively; I xx , I yy , I zz is the moment of inertia; m For drone quality; L is the distance between the motor shaft and the center of the fuselage; g represents the acceleration due to gravity; u 4 , u 5 , u 6 , u 7 Control input for drone 901.

[0031] To sum up, the present invention integrates communication technology, prediction technology, data fusion technology, automatic control technology and artificial intelligence technology, and designs an intelligent control system for a dual-purpose land and air tunnel inspection robot. Through the combination of an integrated operation control model 2, a split operation control model 3, a perception fusion module 1, a path planning module 4, a network communication module 6, an obstacle observation and prediction module 7, an intelligent control module 5 and a charging module 8, the land and air dual-purpose tunnel inspection robot 9 composed of a drone 901 and an unmanned vehicle 902 is endowed with high reliability, strong security, strong autonomy and high intelligence, which has the beneficial effect of improving the safety and reliability of the tunnel inspection robot.

[0032] The beneficial effects of the present invention are described in detail in combination with the above technical solutions, as follows: (1) The present invention integrates automatic control technology and artificial intelligence technology, and uses the tangent network model of the obstacle observation and prediction module 7 and the A-star algorithm of the path planning module 4 to plan the flight path of the drone 901, the driving path of the unmanned vehicle 902, and the landing path from the drone 901 to the unmanned vehicle 902. The planned path is reasonable and feasible, and can achieve the effect of effective obstacle avoidance.

[0033] (2) The present invention integrates data fusion technology and artificial intelligence technology, and uses the tangent network model of the obstacle observation and prediction module 7 as a data fusion model to process and apply the data information of the camera 101, radar 102, laser radar 103, ultrasonic sensor 104, inertial measurement sensor 105, and GPS positioning sensor 106 carried by the drone 901 and the unmanned vehicle 902.

[0034] (3) The present invention integrates communication technology and enables the UAV 901 and the unmanned vehicle 902 to communicate with each other through the network communication module 6; when the UAV 901 and the unmanned vehicle 902 are operated separately, the unmanned vehicle 902 sends position information to the UAV 901, effectively guiding the UAV 901 to land.

[0035] (4) The present invention integrates prediction technology and artificial intelligence technology. The obstacle observation and prediction module 7 uses the tangent network model to observe and predict the position and state information of external obstacles, observe and predict obstacles, and transmit the corresponding obstacle information to the path planning module 4, so that the planned path is optimal and obstacles can be effectively avoided.

[0036] (5) The present invention integrates prediction technology and artificial intelligence technology. The path planning module 4 uses the tangent network to predict the driving trajectory of the unmanned vehicle, updates the path planning end position in real time, and combines the A-star algorithm to plan the landing path of the unmanned vehicle 902 by the unmanned aerial vehicle 901 in real time.

[0037] (6) The present invention integrates automatic control technology and artificial intelligence technology, adopts a refined control strategy, and implements a refined control strategy by using a tangent network controller through the intelligent control module 5; the tangent network controller controls the unmanned vehicle 902, the unmanned aerial vehicle 901, and the unmanned vehicle 902 and the unmanned aerial vehicle 901 when they need to run in an integrated manner to track the optimal driving path planned by the path planning module 4.

[0038] (7) The tangent network model described in the present invention has strong nonlinear approximation, fast convergence, strong generalization ability, "self-learning ability", robustness and anti-interference performance, and can be well applied as an approximation model and classification model.

[0039] It can be understood that the above specific description of the present invention is only used to illustrate the present invention and is not limited to the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that the present invention can still be modified or replaced by equivalents to achieve the same technical effects; as long as the use requirements are met, they are within the protection scope of the present invention.

Claims

1. An intelligent control system for a land-air dual-purpose tunnel inspection robot, characterized in that: The dual-purpose tunnel inspection robot includes a drone and an unmanned vehicle, which are used in conjunction with each other. The coordinated use of the drone and the unmanned vehicle includes controlling the drone and the unmanned vehicle to operate in an integrated manner or controlling the drone and the unmanned vehicle to operate in a separate manner. Controlling the drone and the unmanned vehicle to operate in an integrated manner: when the unmanned vehicle encounters a difficult section of road, the drone is controlled to land on the unmanned vehicle and fix between the unmanned vehicle, and the unmanned vehicle is driven by the drone to fly over the section. Controlling the drone and the unmanned vehicle to operate in a separate manner: when on a normal section of road, the drone and the unmanned vehicle are controlled to operate independently and conduct joint inspections. The intelligent control system includes a perception fusion module, a path planning module, a network communication module, an obstacle observation and prediction module, and an intelligent control module. The perception fusion module is respectively arranged on the UAV and the unmanned vehicle. The perception fusion module is used to detect the external state information of the UAV and the unmanned vehicle. The perception fusion module includes a camera, a radar, a laser radar, an ultrasonic sensor, an inertial measurement sensor, a GPS positioning sensor, a smoke alarm sensor, and a toxic gas detection sensor; the network communication module is used for mutual communication between the UAV and the unmanned vehicle; On normal roads, the path planning module plans the driving paths of the unmanned vehicle and the unmanned aerial vehicle respectively; at the same time, the obstacle observation and prediction module observes and predicts the position and status information of external obstacles, and feeds back the observed and predicted position and status information of external obstacles to the path planning module, so as to cooperate with the path planning module to plan the optimal driving paths for the unmanned vehicle and the unmanned aerial vehicle respectively; When the unmanned vehicle encounters a difficult-to-drive section of road, the path planning module plans the optimal driving path for the drone to land on the unmanned vehicle. After the drone lands on the unmanned vehicle, the drone and the unmanned vehicle are fixed by magnetic attraction to complete the control of the integrated operation of the drone and the unmanned vehicle; according to the optimal driving paths for the unmanned vehicle, drone, and the integration of unmanned vehicle and drone provided by the path planning module, the intelligent control module controls the unmanned vehicle, drone, and the unmanned vehicle and drone when they need to operate in an integrated manner to track the optimal driving paths.

2. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 1 is characterized in that: The operation of the integration process of the UAV and the unmanned vehicle is controlled by the integrated operation control model. The expression of the integrated operation control model is as follows: ; Among them, for the UAV operation control model z i , u i It is expressed as: , ; For the unmanned vehicle operation control model z i , u i It is expressed as: , ;in, A , B , C is the coefficient matrix; X is the horizontal axis, Y is the vertical axis; T is the transposition symbol.

3. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 1 is characterized in that: The path planning module includes an A-star algorithm. The path planning module uses the A-star algorithm in conjunction with the obstacle observation and prediction module to plan the optimal driving path for the unmanned vehicle and the unmanned aerial vehicle respectively. The A-star algorithm is used for global planning. The obstacle observation and prediction module includes a tangent network model. The obstacle observation and prediction module uses the tangent network model to observe and predict the position and status information of external obstacles. The input of the tangent network model is the external status information detected by the camera, radar, lidar, ultrasonic sensor, inertial measurement sensor, GPS positioning sensor, smoke alarm sensor, and toxic gas detection sensor. The output of the tangent network model is the obstacle position and status information. The obstacle observation and prediction module feeds back the obstacle position and status information to the A-star algorithm, and the obstacle position and status information is used as the constraint condition of the A-star algorithm.

4. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 3 is characterized by: The path planning module also includes a tangent network model; when the unmanned vehicle encounters a difficult-to-drive section of road, the path planning module plans the optimal driving path for the drone to land on the unmanned vehicle. After the drone lands on the unmanned vehicle, the drone and the unmanned vehicle are fixed by magnetic attraction to complete the integrated operation of the drone and the unmanned vehicle, specifically including: the unmanned vehicle transmits its own position information to the drone in real time, the path planning module uses the tangent network model to predict the driving trajectory of the unmanned vehicle, the input of the tangent network model is the historical position information of the unmanned vehicle, and the output is the predicted position of the unmanned vehicle. The drone path planning terminal position is updated in real time, and the path for the drone to land on the unmanned vehicle is planned in real time in combination with the A-star algorithm.

5. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 1 is characterized by: The intelligent control system also includes a charging module, which is installed on the unmanned vehicle. When the drone needs to be charged, the path planning module plans the optimal driving path for the drone to land on the unmanned vehicle to complete the integrated operation of the drone and the unmanned vehicle. After the drone and the unmanned vehicle are integrated and fixed, the charging module charges the drone.

6. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 4 is characterized by: The tangent network model takes the following form: ； in, is the tangent network model weight, , x is the network input; the above formula is written as t The general form of the dynamics equation of a dimensional system is: ； in, is the tangent network model weight, x For network input, λ To expand the power, q is the number of items and , ε is the approximation error, and T is the transposed sign.

7. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 6 is characterized in that: The tangent network model optimizes its parameters using a learning algorithm, and the learning algorithm adopts any one of a recursive least squares algorithm, a back propagation algorithm, and a wake-sleep algorithm.

8. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 1 is characterized by: The intelligent control module adopts a tangent network controller, and the form of the tangent network controller is as follows: ； Among them, the controller u 1 Input is ; ; r 1 is the optimal driving path for the integrated operation of the UAV and the unmanned vehicle, which is provided by the path planning module; y 1 is the integrated output trajectory of the UAV and the unmanned vehicle; is the weight of the tangent network model; x Input for the network; λ To expand the power; q is the number of items, ; T is the transposition symbol; ； Among them, the controller u 2 Input is ; ; r 2 is the optimal driving path of the unmanned vehicle, which is provided by the path planning module; y 2 is the output trajectory of the unmanned vehicle; is the weight of the tangent network model; x Input for the network; λ To expand the power; q is the number of items, ; T is the transposition symbol; ； Among them, the controller u 3 Input is ; ; r 3 is the optimal driving path of the UAV, which is provided by the path planning module; y 3 is the output trajectory of the drone; is the weight of the tangent network model; x Input for the network; λ To expand the power; q is the number of items, ; T is the transposition symbol.

9. The intelligent control system of the land-air dual-purpose tunnel inspection robot according to claim 1 is characterized by: The split operation control model is used to control the split operation of the UAV and the unmanned vehicle. The split operation control model includes an independent operation control model for the unmanned vehicle and an independent operation control model for the UAV. The specific control model of the unmanned vehicle independent operation is as follows: ； in, x , y for location; v x , v y They are x axis, y Speed ​​in the axis direction; θ is the yaw angle; ω is the angular velocity of rotation; The specific control model of independent operation of UAV is as follows: ； in, x , y , z is the spatial position; φ , θ , ψ They represent the roll angle, pitch angle and yaw angle respectively; I xx , I yy , I zz is the moment of inertia; m For drone quality; L is the distance between the motor shaft and the center of the fuselage; g represents the acceleration due to gravity; u 4. u 5. u 6. u 7 is the drone control input.

Images