Composite wheel track triangle wheel omni-directional chassis and dynamic weight trajectory planning method

Abstract

This invention discloses a composite tracked triangular wheel omnidirectional chassis and a dynamic weighted trajectory planning method, relating to the field of autonomous navigation technology. A visual perception module processes acquired terrain images and obstacle information and inputs them into a perception network, outputting terrain complexity coefficients and obstacle 3D coordinates. A heterogeneous motion constraint model of the composite tracked triangular wheel chassis is constructed to determine kinematic boundaries. A terrain-chassis dual-drive dynamic weighted evolution model is established, introducing coupling factors to dynamically allocate four-dimensional weights. A dynamic velocity window is generated through a two-layer strategy of illegal velocity pruning and artificial potential field obstacle avoidance. The trajectory is predicted based on an improved DWA algorithm, and the trajectory score is calculated using four-dimensional weights. The optimal trajectory is output, and the motion state is fed back to form a closed loop. This invention uses the aforementioned composite tracked triangular wheel omnidirectional chassis and dynamic weighted trajectory planning method to improve the problems of single chassis performance and poor algorithm-hardware compatibility.

Description

Technical Field

[0001] This invention relates to the field of autonomous navigation technology, and in particular to a composite tracked triangular wheel omnidirectional chassis and a dynamic weighted trajectory planning method. Background Technology

[0002] Quadruped chassis have become a research focus in the field of mobile chassis due to their excellent terrain adaptability. However, the quality of their motion performance depends primarily on the synergistic effect of the chassis structure and trajectory planning algorithm. At present, there are still several core shortcomings in related technologies that need to be addressed: The chassis structure design has limitations: traditional four-legged chassis mostly adopt pure track, pure wheel or ordinary round wheel structure. Pure track has strong traction but poor omnidirectional mobility. Pure wheel has good omnidirectionality but is easy to slip on complex terrain. Round wheel has small contact area and insufficient grip, and cannot meet the needs of omnidirectional mobility and strong traction in mixed terrain. Poor algorithm and chassis adaptability: Traditional trajectory planning algorithms such as the Dynamic Window Approach (DWA) adopt a fixed weight design and do not dynamically adjust in combination with chassis structural characteristics. The obstacle avoidance strategy is only based on distance threshold judgment, and the integration with terrain perception and chassis constraints is low, resulting in the trajectory being disconnected from the actual motion state of the chassis. Under mixed terrain, the trajectory deviation is large and the risk of mechanical failure is high. Insufficient synergy between perception and planning: terrain perception and trajectory planning are mostly designed separately. The perception results are only used for simple shape classification and are not deeply coupled with core links such as weight evolution, velocity sampling, and trajectory evaluation, which cannot give full play to the advantages of the chassis structure. Real-time performance is difficult to meet actual needs: The complex terrain perception network has a large number of parameters and takes a long time to infer, making it impossible to run in real time on the main control chip at the edge of the chassis, and it is difficult to adapt to the dynamic motion requirements of the four-legged chassis.

[0003] To address the aforementioned pain points, this invention designs a composite omnidirectional chassis with a triangular wheel and a track, integrating the triple advantages of the large contact area of ​​the triangular wheel, the omnidirectional movement of the track, and the strong traction of the track. It is also equipped with an improved DWA trajectory planning method with terrain-chassis dual drive to achieve deep coupling of perception, constraint, planning, and obstacle avoidance. Summary of the Invention

[0004] The purpose of this invention is to provide a composite tracked triangular wheel omnidirectional chassis and a dynamic weighted trajectory planning method. It provides a four-legged chassis that takes into account omnidirectional mobility, traction in complex terrain, and motion stability, as well as a formulaic and quantifiable improved DWA trajectory planning method adapted to the chassis depth. Through chassis structure innovation and systematic optimization of the mathematical model of the algorithm, the trajectory accuracy, obstacle avoidance ability, and real-time performance in mixed terrain are improved, while reducing the risk of mechanical failure and energy consumption.

[0005] To achieve the above objectives, the present invention provides a composite tracked triangular wheel omnidirectional chassis, including a chassis, a control box and a power supply compartment disposed in the middle of the chassis, the power supply compartment being electrically connected to the control box, a drive motor disposed at each of the four corners of the chassis, each drive motor being connected to a composite tracked triangular wheel, a vision integrator disposed between the two drive motors on the front side of the control box, and the control box being electrically connected to the drive motors and the vision integrator.

[0006] Preferably, the composite track triangular wheel includes a track, with a plurality of tracks arranged on the outer surface of the track, a support frame inside the track, a drive wheel at the top of the support frame, a triangular support connecting the drive wheel and the support frame, a load-bearing wheel set at the bottom of the support frame, tension wheels at both ends of the support frame, and a rack on the inner surface of the track, which meshes with the drive wheel.

[0007] Preferably, the rollers are arranged at a 45-degree angle, the spacing between the rollers is 5-8mm, the rollers are made of reinforced polyoxymethylene material, and the tracks are made of neoprene rubber material.

[0008] Preferably, the vision integrator includes a camera and a lidar, which are electrically connected to the control box.

[0009] Preferably, the control box is equipped with a visual perception module, a dual-output perception network module, a heterogeneous motion constraint module, a dynamic weight evolution module, a two-level obstacle avoidance optimization module, an improved DWA trajectory planning module, and a motion state feedback module. The visual perception module processes the acquired terrain images and obstacle information and inputs them into the dual-output perception network module; the dual-output perception network outputs a terrain complexity coefficient based on the terrain images and obstacle information. The system receives obstacle coordinates and motion state feedback; a heterogeneous motion constraint module generates chassis physical boundary constraints based on terrain complexity coefficients and motion state feedback; a dynamic weight evolution module dynamically allocates four-dimensional weights based on terrain and chassis state; a two-level obstacle avoidance optimization module first prunes illegal velocities to protect the chassis, and then avoids obstacles using an artificial potential field method; an improved DWA trajectory planning module samples velocity combinations, predicts multiple trajectories in parallel, and selects the optimal one; and a motion state feedback module returns motion parameters and updates the terrain complexity coefficients. It forms a closed loop with the boundary constraints.

[0010] To achieve the above objectives, the present invention also provides a dynamic weighted trajectory planning method for a composite tracked triangular wheel omnidirectional chassis, comprising the following steps: S1. The visual perception module processes the terrain image and obstacle information acquired by the visual integrator and inputs it into the dual-output perception network, outputting a terrain complexity coefficient of 0 to 1. and the three-dimensional coordinates of the obstacle; S2. Construct a heterogeneous motion constraint model for a composite tracked triangular wheel chassis and determine the longitudinal asymmetric velocity threshold, the lateral asymmetric velocity threshold, the steering angular velocity constraint, and the terrain adaptation kinematic boundary. S3. Establish a terrain-chassis dual-drive dynamic weight evolution model and introduce coupling factors. Dynamically allocate weights for four dimensions: heading angle, speed, obstacle avoidance, and chassis adaptability. S4. Optimize the generation of dynamic velocity windows through a two-layer strategy of illegal velocity pruning and obstacle avoidance using artificial potential field method; S5. Based on the improved DWA algorithm, the trajectory is predicted, the trajectory score is calculated through the four-dimensional dynamic weight evaluation function, the optimal trajectory is output and the motion state is fed back to form a closed loop.

[0011] Preferably, the process of constructing the heterogeneous motion constraint model in S2 is as follows: S21. Based on the characteristics of the chassis structure, construct an asymmetric motion constraint mathematical model, with the core constraint formula as follows: ; in, The lateral translation speed of the chassis, i.e., the speed of movement along the left and right directions of the chassis, is mainly achieved by the lateral force characteristics of the wheels. The longitudinal travel speed of the chassis, i.e., the speed of movement along the front-to-back direction of the chassis, is mainly achieved by the traction force of the tracks and drive wheels. The chassis steering angular velocity is the rate at which the chassis rotates around its own center. The terrain complexity coefficient is output in real time by a dual-output perception network, with a value ranging from 0 to 1, and is used to reflect the current ruggedness of the road surface. The speed of the track drive wheel is used to directly monitor and constrain the physical motion boundaries of the motor; S22, Based on maximum acceleration The dynamic velocity window is generated, and the boundary formula is: ; in, This represents the current real-time lateral translation speed of the chassis, which is the actual execution speed at the previous sampling time. This represents the current real-time longitudinal speed of the chassis. The current angular velocity, For the sampling time step, For the maximum angular velocity, The lateral physical velocity limit under the current terrain constraints, i.e., according to S21 The boundary values ​​are defined as [-0.8, 0.8] or [0.3, 0.8]. The longitudinal physical velocity limit under the current terrain constraints is the boundary value [-0.5, 1.2] defined in S21. The maximum steering angular velocity limit under the current terrain constraints is defined in S21. .

[0012] Preferably, the S3 model establishes a terrain-chassis dual-drive dynamic weight evolution model by introducing a coupling factor. , To minimize the value, avoid a denominator of 0 and , The contact area coefficient between the chassis and the ground is... ≥1.5, combined and Dynamically allocate heading angle ,speed Obstacle avoidance Chassis adaptability The four-dimensional weights and The allocation formula is as follows: ( )= ; Preferably, the specific process of S4 is as follows: S41. Illegal speed pruning is a form of removal. hour , Speed ​​combinations; S42. Artificial potential field obstacle avoidance is based on the target point. For gravitational fields, obstacles Given a repulsive field, calculate the potential field force. Eliminate velocity combinations where the repulsive force exceeds the threshold. S43, Gravity of a Gravitational Field The calculation formula is as follows: ; in, It is the gravitational gain coefficient and , Let x be the x-coordinate of the target point. The x-coordinate represents the current position of the chassis. The ordinate of the target point, The vertical coordinate of the current position of the chassis; S44. The formula for calculating the repulsive force of a repulsive field is as follows: ; in, The repulsive gain coefficient and , The range of the obstacle's influence and , This is the Euclidean distance from the chassis to the obstacle.

[0013] Preferably, the specific process of S5 is as follows: S51. Predicting the trajectory in the next 1 second based on kinematic formulas. : ; in, For the predicted time The chassis's global horizontal coordinate. For the predicted time The chassis's global vertical coordinates, For the predicted time Chassis heading angle, For the predicted time The cumulative state value of the track drive wheel. For the previous prediction time The chassis's global horizontal coordinate. For the previous prediction time The chassis's global vertical coordinates, For the previous prediction time Chassis heading angle, For the previous prediction time The cumulative state value of the track drive wheels; S52. Calculate the trajectory score using the following formula: ; in, Score for heading angle. Score points for speed. To score points by avoiding obstacles, The motor efficiency coefficient. To determine the chassis adaptability score, the optimal trajectory is selected for output, and the actual motion state is updated in response. Together with the four-dimensional weights, they form a closed loop; S53. Real-time monitoring of execution deviations and acquisition of motor status data, including the actual chassis speed. Using motor current characteristics as physical layer feedback, the heterogeneous motion constraint module is input to the terrain complexity coefficient. Perform real-time deviation correction and cycle through S1-S5 until the chassis reaches the preset target point range.

[0014] Therefore, the composite tracked triangular wheel omnidirectional chassis and dynamic weighted trajectory planning method described above have the following advantages compared with the prior art: 1. The four-legged chassis of this application integrates multiple performance characteristics. The composite wheel track triangular wheel integrates three core characteristics. The contact area of ​​a single wheel is increased by 50% compared with the traditional round wheel, the grip is enhanced by 40%, the traction of complex terrain is improved by 60%, and the omnidirectional movement response time on flat terrain is optimized by 85%, thus improving the pain point of single chassis performance. 2. The algorithm in this application is accurate and quantifiable. The improved DWA algorithm is constructed through a full-process mathematical formula, replacing the traditional fuzzy description. The chassis heading deviation angle under mixed terrain is ≤2°, which is more than 90% lower than the traditional fixed-weight DWA algorithm. It has strong obstacle avoidance and scene adaptability. The two-level obstacle avoidance strategy takes into account both chassis protection and obstacle avoidance effect. The obstacle collision avoidance success rate can reach more than 99.5%. The chassis can cope with various mixed terrains and is suitable for multiple scenarios. The algorithm framework can be ported to other wheel-tracked composite chassis, and its application scalability is wide. 3. Significantly enhanced motion stability: Through heterogeneous motion constraints and dual-level obstacle avoidance pruning, mechanical failures are avoided at the planning level. The failure rate of sideslip and tooth loss in complex terrain is reduced by more than 90%, the average motor current fluctuation is ≤2%, energy consumption is reduced by 35%, and the range is increased by 40%. Excellent real-time performance and engineering capabilities: the inference time of the lightweight perception network is ≤10ms, the overall algorithm planning cycle is ≤20ms, and it can run in real time on the edge main control chip. The chassis adopts mature materials and standardized processes, the structure is easy to process, and the formulaic design of the algorithm facilitates engineering implementation and debugging.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] Figure 1 This is an overall structural diagram of the composite wheel track triangular wheel omnidirectional chassis of the present invention; Figure 2 This is a structural diagram of the composite wheel track triangular wheel omnidirectional chassis of the present invention; Figure 3 This is a block diagram of the control module of the composite tracked triangular wheel omnidirectional chassis of the present invention; Figure 4 This is an overall flowchart of the dynamic weighted trajectory planning method for the composite wheel track triangular wheel omnidirectional chassis of the present invention; Figure 5 This is a comparison chart of heading deviations for the dynamic weighted trajectory planning method of the composite McLaren track triangular wheel omnidirectional chassis of the present invention; Figure 6 This is a comparison chart of the motion performance of the dynamic weighted trajectory planning method for the composite wheel track triangular wheel omnidirectional chassis of the present invention.

[0017] Figure Labels 1. Chassis; 2. Control box; 3. Power supply compartment; 4. Drive motor; 5. Composite track wheel and triangular wheel; 6. Vision integrator; 61. Camera; 62. LiDAR; 51. Track; 52. Track wheel; 53. Support frame; 54. Drive wheel; 55. Triangular support; 56. Load-bearing wheel set; 57. Tensioner wheel; 58. Rack and pinion. Detailed Implementation

[0018] In the description of this invention, it should be noted that the terms "upper," "lower," "inner," "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product of this invention is usually placed when in use. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0019] Example like Figures 1-3 As shown, the composite tracked triangular wheel omnidirectional chassis of the present invention includes a chassis 1, a control box 2 and a power supply compartment 3 are arranged in the middle of the chassis 1, the power supply compartment 3 is electrically connected to the control box 2, a drive motor 4 is arranged at each of the four corners of the chassis 1, and each drive motor 4 is connected to a composite tracked triangular wheel 5. A vision integrator 6 is arranged between the two drive motors 4 on the front side of the control box 2, and the control box 2 is electrically connected to the drive motors 4 and the vision integrator 6.

[0020] The composite track triangular wheel 5 includes a track 51, with a plurality of tracks 52 arranged on the outer surface of the track 51. A support frame 53 is provided inside the track 51, a drive wheel 54 is provided on the top of the support frame 53, a triangular support 55 is connected between the drive wheel 54 and the support frame 53, a load-bearing wheel set 56 is provided at the bottom of the support frame 53, tension wheels 57 are provided at both ends of the support frame 53, and a rack 58 is provided on the inner surface of the track 51, which meshes with the drive wheel 54.

[0021] All the rollers 52 are arranged at a 45-degree angle, with a spacing of 5-8 mm between them. The rollers 52 are made of reinforced polyoxymethylene material, and the tracks 51 are made of neoprene rubber.

[0022] The vision integrator 6 includes a camera 61 and a lidar 62, which are electrically connected to the control box 2.

[0023] The control box 2 is equipped with a visual perception module, a dual-output perception network module, a heterogeneous motion constraint module, a dynamic weight evolution module, a two-level obstacle avoidance optimization module, an improved DWA trajectory planning module, and a motion state feedback module. The visual perception module processes the collected terrain images and obstacle information and inputs them into the dual-output perception network module; the dual-output perception network outputs the terrain complexity coefficient based on the terrain images and obstacle information. The system receives obstacle coordinates and motion state feedback; a heterogeneous motion constraint module generates chassis physical boundary constraints based on terrain complexity coefficients and motion state feedback; a dynamic weight evolution module dynamically allocates four-dimensional weights based on terrain and chassis state; a two-level obstacle avoidance optimization module first prunes illegal velocities to protect the chassis, and then avoids obstacles using an artificial potential field method; an improved DWA trajectory planning module samples velocity combinations, predicts multiple trajectories in parallel, and selects the optimal one; and a motion state feedback module returns motion parameters and updates the terrain complexity coefficients. It forms a closed loop with the boundary constraints.

[0024] like Figures 4-6 As shown, the composite tracked triangular wheel omnidirectional chassis and dynamic weighted trajectory planning method of the present invention include the following steps: S1. The visual perception module processes the terrain image and obstacle information acquired by the visual integrator and inputs it into the dual-output perception network (TerrainPerceptionNet), outputting a terrain complexity coefficient of 0 to 1. and the three-dimensional coordinates of the obstacle; S2. Construct a heterogeneous motion constraint model for a composite tracked triangular wheel chassis and determine the longitudinal asymmetric velocity threshold, the lateral asymmetric velocity threshold, the steering angular velocity constraint, and the terrain adaptation kinematic boundary. The process of constructing the heterogeneous motion constraint model is as follows: S21. Based on the characteristics of the chassis structure, construct an asymmetric motion constraint mathematical model, with the core constraint formula as follows: ; in, The lateral translation speed of the chassis, i.e., the speed of movement along the left and right directions of the chassis, is mainly achieved by the lateral force characteristics of the wheels. The longitudinal travel speed of the chassis, i.e., the speed of movement along the front-to-back direction of the chassis, is mainly achieved by the traction force of the tracks and drive wheels. The chassis steering angular velocity is the rate at which the chassis rotates around its own center. The terrain complexity coefficient is output in real time by a dual-output perception network, with a value ranging from 0 to 1, and is used to reflect the current ruggedness of the road surface. The speed of the track drive wheel is used to directly monitor and constrain the physical motion boundaries of the motor; S22, Based on maximum acceleration The dynamic velocity window is generated, and the boundary formula is: ; in, This represents the current real-time lateral translation speed of the chassis, which is the actual execution speed at the previous sampling time. This represents the current real-time longitudinal speed of the chassis. The current angular velocity, For the sampling time step, For the maximum angular velocity, The lateral physical velocity limit under the current terrain constraints, i.e., according to S21 The boundary values ​​are defined as [-0.8, 0.8] or [0.3, 0.8]. The longitudinal physical velocity limit under the current terrain constraints is the boundary value [-0.5, 1.2] defined in S21. The maximum steering angular velocity limit under the current terrain constraints is defined in S21. ; S3. Establish a terrain-chassis dual-drive dynamic weight evolution model and introduce coupling factors. Dynamically allocate weights for four dimensions: heading angle, speed, obstacle avoidance, and chassis adaptability. Establish a terrain-chassis dual-drive dynamic weight evolution model and introduce coupling factors. , To minimize the value, avoid a denominator of 0 and , The contact area coefficient between the chassis and the ground is... ≥1.5, combined and Dynamically allocate heading angle ,speed Obstacle avoidance Chassis adaptability The four-dimensional weights and The allocation formula is as follows: ( )= ; S4. Optimize the generation of dynamic velocity windows through a two-layer strategy of illegal velocity pruning and obstacle avoidance using artificial potential field method; S41. Illegal speed pruning is a form of removal. hour , Speed ​​combinations; S42. Artificial potential field obstacle avoidance is based on the target point. For gravitational fields, obstacles Given a repulsive field, calculate the potential field force. Eliminate velocity combinations where the repulsive force exceeds the threshold. S43, Gravity of a Gravitational Field The calculation formula is as follows: ; in, It is the gravitational gain coefficient and , Let x be the x-coordinate of the target point. The x-coordinate represents the current position of the chassis. The ordinate of the target point, The vertical coordinate of the current position of the chassis; S44. The formula for calculating the repulsive force of a repulsive field is as follows: ; in, The repulsive gain coefficient and , This refers to the radius of influence of the obstacle (meaning the repulsive force only takes effect when the obstacle enters within 1.8 meters of the chassis; outside this range, the repulsive force is zero). , The distance from the chassis to the obstacle is the Euclidean distance. S5. Based on the improved DWA algorithm, the trajectory is predicted, the trajectory score is calculated through the four-dimensional dynamic weight evaluation function, the optimal trajectory is output and the motion state is fed back to form a closed loop. S51. Predicting the trajectory in the next 1 second based on kinematic formulas. : ; in, For the predicted time The chassis's global horizontal coordinate. For the predicted time The chassis's global vertical coordinates, For the predicted time Chassis heading angle, For the predicted time The cumulative state value of the track drive wheel. For the previous prediction time The chassis's global horizontal coordinate. For the previous prediction time The chassis's global vertical coordinates, For the previous prediction time Chassis heading angle, For the previous prediction time The cumulative state value of the track drive wheels; S52. Calculate the trajectory score using the following formula: ; in, Score for heading angle. Score points for speed. To score points by avoiding obstacles, The motor efficiency coefficient. To determine the chassis adaptability score, the optimal trajectory is selected for output, and the actual motion state is updated in response. Together with the four-dimensional weights, they form a closed loop; S53. Real-time monitoring of execution deviations and acquisition of motor status data, including the actual chassis speed. Using motor current characteristics as physical layer feedback, the heterogeneous motion constraint module is input to the terrain complexity coefficient. Perform real-time deviation correction and cycle through S1-S5 until the chassis reaches the preset target point range.

[0025] In the specific implementation process, regarding the structural design of the composite tracked triangular wheel four-legged chassis: The four sets of composite track rollers are an integrated structure of equilateral triangular wheel bodies and omnidirectional track rollers. The track side length is 80-120mm, and the track is evenly distributed with alternating orthogonal track rollers at 45° angles. The track roller diameter is 15-20mm, and the spacing between adjacent track rollers is 5-8mm. The track and track rollers are rigidly fitted, and the contact area between a single track roller and the ground is ≥12cm². 2 The track and wheel drive efficiency is ≥90%, integrating the triple characteristics of large contact area, omnidirectional movement, and strong traction; The vision integrator integrates a 2-megapixel industrial camera (sampling frame rate 30fps, image resolution 1920×1080) and a 16-line LiDAR (maximum detection distance 10m, ranging accuracy ±2cm), and is mounted on the top of a bracket on the chassis. It collects terrain visual images and 3D coordinates of obstacles to provide accurate environmental information for trajectory planning. The chassis is constructed from four 150-200mm thick 6061 aluminum alloy rods. The power supply compartment, control box, and drive motor are located on the chassis. The drive motor is connected to the composite track triangular wheel via a connecting shaft. The connection part is reinforced with a thickness of ≥3mm, forming a four-legged symmetrical omnidirectional mobile chassis. The chassis has an overall load capacity of ≥5kg, a maximum obstacle crossing height of ≥30mm, and a turning angular velocity of ≤15° / s, ensuring the structural stability of the whole machine and adaptability to various terrains.

[0026] The trajectory optimization method for a four-legged chassis with composite tracked wheels and triangular wheels is as follows: Using a four-legged chassis equipped with a composite tracked triangular wheel omnidirectional chassis as the carrier, the main control chip is a Jetson OrinNX with 8GB of memory. The core parameters of the chassis are longitudinal speed -0.5~1.2m / s, lateral speed -0.8~0.8m / s, and track rotation angular velocity ≤20r / min. The trajectory planning algorithm on this main control chip has an overall planning cycle of ≤20ms, which is compatible with the 30fps sampling frame rate of the vision integrator. It completes the entire process operation according to the closed loop process of perception-constraint-planning-obstacle avoidance-execution-feedback, realizing stable and precise movement of the chassis in flat, transitional, and complex mixed terrain.

[0027] Therefore, this invention adopts the composite tracked triangular wheel omnidirectional chassis and dynamic weighted trajectory planning method described above to provide a four-legged chassis that takes into account omnidirectional mobility, traction in complex terrain, and motion stability, as well as a formulaic and quantifiable improved DWA trajectory planning method adapted to the chassis depth. Through chassis structure innovation and systematic optimization of the mathematical model of the algorithm, the trajectory accuracy, obstacle avoidance ability, and real-time performance in mixed terrain are improved, while reducing the risk of mechanical failure and energy consumption.

[0028] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A composite tracked triangular wheel omnidirectional chassis, characterized in that, The system includes a chassis, with a control box and a power supply compartment located in the middle of the chassis. The power supply compartment is electrically connected to the control box. Each of the four corners of the chassis has a drive motor, and each drive motor is connected to a composite track triangular wheel. A vision integrator is located between the two drive motors on the front side of the control box. The control box is electrically connected to the drive motors and the vision integrator.

2. The composite tracked triangular wheel omnidirectional chassis according to claim 1, characterized in that: The composite track triangular wheel includes a track, with several tracks arranged on the outer surface of the track. A support frame is installed inside the track, a drive wheel is installed on the top of the support frame, a triangular support is connected between the drive wheel and the support frame, a load-bearing wheel set is installed at the bottom of the support frame, tension wheels are installed at both ends of the support frame, and a rack is installed on the inner surface of the track, which meshes with the drive wheel.

3. The composite tracked triangular wheel omnidirectional chassis according to claim 2, characterized in that: The wheels are all arranged at a 45-degree angle, with a spacing of 5-8mm between them. The wheels are made of reinforced polyoxymethylene, and the tracks are made of neoprene.

4. The composite tracked triangular wheel omnidirectional chassis according to claim 1, characterized in that: The vision integrator includes a camera and a lidar unit, which are electrically connected to the control box.

5. The composite tracked triangular wheel omnidirectional chassis according to claim 4, characterized in that: The control box contains a visual perception module, a dual-output perception network module, a heterogeneous motion constraint module, a dynamic weight evolution module, a two-level obstacle avoidance optimization module, an improved DWA trajectory planning module, and a motion state feedback module. The visual perception module processes the acquired terrain images and obstacle information and inputs them into the dual-output perception network module. The dual-output perception network outputs a terrain complexity coefficient based on the terrain images and obstacle information. The system receives obstacle coordinates and motion state feedback; the heterogeneous motion constraint module generates chassis physical boundary constraints based on terrain complexity coefficient and motion state feedback; the dynamic weight evolution module dynamically allocates four-dimensional weights based on terrain and chassis state. The dual-level obstacle avoidance optimization module first prunes illegal speeds to protect the chassis, and then avoids obstacles using an artificial potential field method; the DWA trajectory planning module is improved to sample speed combinations, predict multiple trajectories in parallel, and select the optimal one. The motion state feedback module sends back motion parameters and updates the terrain complexity coefficient. It forms a closed loop with the boundary constraints.

6. A dynamic weighted trajectory planning method for a composite tracked triangular wheel omnidirectional chassis, characterized in that: The method of using a composite tracked triangular wheel omnidirectional chassis as described in any one of claims 1-5 includes the following steps: S1. The visual perception module processes the terrain image and obstacle information acquired by the visual integrator and inputs it into the dual-output perception network, outputting a terrain complexity coefficient of 0 to 1. and the three-dimensional coordinates of the obstacle; S2. Construct a heterogeneous motion constraint model for a composite tracked triangular wheel chassis and determine the longitudinal asymmetric velocity threshold, the lateral asymmetric velocity threshold, the steering angular velocity constraint, and the terrain adaptation kinematic boundary. S3. Establish a terrain-chassis dual-drive dynamic weight evolution model and introduce coupling factors. Dynamically allocate weights for four dimensions: heading angle, speed, obstacle avoidance, and chassis adaptability. S4. Optimize the generation of dynamic velocity windows through a two-layer strategy of illegal velocity pruning and obstacle avoidance using artificial potential field method; S5. Based on the improved DWA algorithm, the trajectory is predicted, the trajectory score is calculated through the four-dimensional dynamic weight evaluation function, the optimal trajectory is output and the motion state is fed back to form a closed loop.

7. The dynamic weighted trajectory planning method for the composite tracked triangular wheel omnidirectional chassis according to claim 6, characterized in that: The process of constructing the heterogeneous motion constraint model in S2 is as follows: S21. Based on the chassis structural characteristics, construct an asymmetric motion constraint mathematical model, with the core constraint formula as follows: ； in, The lateral translation speed of the chassis, i.e., the speed of movement along the left and right directions of the chassis, is mainly achieved by the lateral force characteristics of the wheels. The longitudinal travel speed of the chassis, i.e., the speed of movement along the front-to-back direction of the chassis, is mainly achieved by the traction force of the tracks and drive wheels. The chassis steering angular velocity is the rate at which the chassis rotates around its own center. The terrain complexity coefficient is output in real time by a dual-output perception network, with a value ranging from 0 to 1, and is used to reflect the current ruggedness of the road surface. The speed of the track drive wheel is used to directly monitor and constrain the physical motion boundaries of the motor; S22, Based on maximum acceleration The dynamic velocity window is generated, and the boundary formula is: ； in, This represents the current real-time lateral translation speed of the chassis, which is the actual execution speed at the previous sampling time. This represents the current real-time longitudinal speed of the chassis. The current angular velocity, For the sampling time step, For the maximum angular velocity, The lateral physical velocity limit under the current terrain constraints, i.e., according to S21 The boundary values ​​are defined as [-0.8, 0.8] or [0.3, 0.8]. The longitudinal physical velocity limit under the current terrain constraints is the boundary value [-0.5, 1.2] defined in S21. The maximum steering angular velocity limit under the current terrain constraints is defined in S21. .

8. The dynamic weighted trajectory planning method for the composite tracked triangular wheel omnidirectional chassis according to claim 7, characterized in that: The S3 model establishes a terrain-chassis dual-drive dynamic weight evolution model and introduces coupling factors. , To minimize the value, avoid a denominator of 0 and , The contact area coefficient between the chassis and the ground is... ≥1.5, combined and Dynamically allocate heading angle ,speed Obstacle avoidance Chassis adaptability The four-dimensional weights and The allocation formula is as follows: ( )= 。 9. The dynamic weighted trajectory planning method for the composite tracked triangular wheel omnidirectional chassis according to claim 8, characterized in that: The specific process of S4 is as follows: S41. Illegal speed pruning is a form of removal. hour , Speed ​​combinations; S42. Artificial potential field obstacle avoidance is based on the target point. For gravitational fields, obstacles For a repulsive field, calculate the potential field force. Eliminate velocity combinations where the repulsive force exceeds the threshold. S43, Gravity of a Gravitational Field The calculation formula is as follows: ； in, It is the gravitational gain coefficient and , Let x be the x-coordinate of the target point. The x-coordinate represents the current position of the chassis. The ordinate of the target point, The vertical coordinate of the current position of the chassis; S44. The formula for calculating the repulsive force of a repulsive field is as follows: ； in, The repulsive gain coefficient and , The range of the obstacle's influence and , This is the Euclidean distance from the chassis to the obstacle.

10. The dynamic weighted trajectory planning method for the composite tracked triangular wheel omnidirectional chassis according to claim 9, characterized in that: The specific process of S5 is as follows: S51. Predicting the trajectory in the next 1 second based on kinematic formulas. : ； in, For the predicted time The global horizontal coordinate of the chassis. For the predicted time The chassis's global vertical coordinates, For the predicted time Chassis heading angle, For the predicted time The cumulative state value of the track drive wheel. For the previous prediction time The global horizontal coordinate of the chassis. For the previous prediction time The chassis's global vertical coordinates, For the previous prediction time Chassis heading angle, For the previous prediction time The cumulative state value of the track drive wheels; S52. Calculate the trajectory score using the following formula: ； in, Score for heading angle. Score points for speed. To score points by avoiding obstacles, The motor efficiency coefficient. To determine the chassis adaptability score, the optimal trajectory is selected for output, and the actual motion state is updated in response. Together with the four-dimensional weights, they form a closed loop; S53. Real-time monitoring of execution deviations and acquisition of motor status data, including the actual chassis speed. Using motor current characteristics as physical layer feedback, the heterogeneous motion constraint module is input to the terrain complexity coefficient. Perform real-time deviation correction and cycle through S1-S5 until the chassis reaches the preset target point range.

Images