Mobile robot obstacle avoidance method and system for high-speed dynamic obstacle

By calculating the minimum distance of the space-time trajectory of high-speed dynamic obstacles in the four-dimensional space-time coordinate system and reinforcing learning to optimize the obstacle avoidance path, the decision-making error problem caused by excessive increase in the position of the obstacle in the prior art is solved, and the mobile robot is improved. Obstacle avoidance safety and efficiency in high-speed dynamic environments.

CN119024839APending Publication Date: 2024-11-26HUAZHONG UNIV OF SCI & TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411122436.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-26

Smart Images

  • Figure CN119024839A_ABST
    Figure CN119024839A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of mobile robot obstacle avoidance, and discloses a mobile robot obstacle avoidance method and system for a high-speed dynamic obstacle, and the method comprises the steps: firstly adding a time axis on the basis of a three-dimensional space coordinate system, building a four-dimensional space-time coordinate system, and carrying out the optimization of the time axis in the space-time coordinate system; and calculating the minimum distance between the space-time trajectory of the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling time so as to overcome the defect that the position increment of the high-speed dynamic obstacle at the adjacent sampling time is far greater than the position increment of the robot in the traditional method when facing the high-speed dynamic obstacle. And the calculated risk assessment distance causes an obstacle avoidance decision error. And finally, on the basis of the obstacle distance reward, adding an additional reward related to the obstacle speed, so that the mobile robot can preferentially avoid the obstacle with the higher speed, and the distance between the mobile robot and the obstacle with the higher speed is as far as possible in the avoidance process, thereby improving the safety of the obstacle avoidance process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of mobile robot obstacle avoidance, and in particular relates to a mobile robot obstacle avoidance method and system facing high-speed dynamic obstacles. Background Art

[0002] During real-time movement, mobile robots can generally obtain the changing positions and postures of obstacles in the surrounding environment through their own sensors, and adjust their own paths to avoid collisions. This technology is crucial for mobile robots to drive safely in complex dynamic environments. Commonly used dynamic obstacle avoidance methods mainly include artificial potential field method, dynamic A* algorithm, fuzzy logic algorithm, etc. The artificial potential field method uses the potential field to define the robot's surrounding environment to plan the robot's obstacle avoidance trajectory; the dynamic A* algorithm considers the movement cost of nodes in a dynamic environment to find obstacle avoidance paths; the fuzzy logic algorithm converts environmental information into fuzzy variables and performs reasoning based on preset fuzzy rules to control the robot to make obstacle avoidance decisions in a dynamic environment.

[0003] The main problems and defects in the prior art include: when calculating the risk assessment distance between the robot and the obstacle, the current dynamic obstacle avoidance method generally only calculates the distance between the robot and the obstacle at the sampling moment through a time-segmented sampling method, and when the movement speed of the dynamic obstacle is much greater than the movement speed of the robot, at adjacent sampling moments, the position increment of the obstacle is much greater than the position increment of the robot, which can easily cause the risk assessment distance at the current sampling moment to be greater than the safety threshold while the risk assessment distance at the next sampling moment is less than the safety threshold, which in turn causes the robot to not take evasive action at the current sampling moment, and after judging the risk at the next sampling moment, the execution time of the subsequent evasive action is very limited, which can easily cause danger. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a mobile robot obstacle avoidance method for high-speed dynamic obstacles. The method uses the spatiotemporal trajectory position information of the high-speed dynamic obstacle in the risk assessment time period to replace the position information at the discrete sampling time. When calculating the risk assessment distance information, the minimum value of the spatiotemporal trajectory of the robot and the high-speed dynamic obstacle in the risk assessment time period is calculated at the sampling time, so that the robot can take evasive action in advance at the sampling time to avoid the high-speed dynamic obstacle. The present invention firstly adds a time axis on the basis of a three-dimensional space coordinate system to establish a four-dimensional space-time coordinate system. By calculating the minimum distance between the space-time trajectory of the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling moment in the space-time coordinate system, the present invention overcomes the deficiency of the traditional method in facing high-speed dynamic obstacles, that is, when the position increment of the high-speed dynamic obstacle at adjacent sampling moments is much larger than the position increment of the robot, the calculated risk assessment distance leads to wrong obstacle avoidance decision-making. Then, reinforcement learning is used to select the actions taken by the robot, and the robot control instructions are directly obtained, thereby reducing the solution time of the instructions. Finally, on the basis of the obstacle distance reward, an additional reward related to the obstacle speed is added, so that the mobile robot can give priority to avoiding obstacles with faster speeds, and stay as far away from obstacles with faster speeds as possible during the avoidance process, thereby improving the safety of the obstacle avoidance process.

[0005] The present invention is implemented in this way: a mobile robot obstacle avoidance method facing high-speed dynamic obstacles comprises:

[0006] Step 1: Obtain the status of obstacles and mobile robots through sensors;

[0007] Step 2: According to the state information, the three-dimensional space coordinate system is expanded based on the time dimension to establish a four-dimensional space-time coordinate system;

[0008] Step 3: In the space-time coordinate system, calculate the minimum distance between the robot and the high-speed dynamic obstacle at the sampling moment in the space-time trajectory corresponding to the risk assessment time period;

[0009] Step 4: Build a reinforcement learning environment for mobile robots to avoid obstacles in dynamic scenes;

[0010] Step 5: Design a reward function based on obstacle speed and distance, target distance, and target orientation to guide the mobile robot to successfully avoid obstacles and reach the target point.

[0011] Furthermore, in step 1, the high-speed dynamic obstacle state obtained by the robot sensor is in, is the coordinate of the ith obstacle in the three-dimensional space coordinate system, is the moving speed of the ith obstacle in the three-dimensional space coordinate system, is the moment corresponding to the current state of the ith obstacle; the state of the robot is S r =[x r ,y r ,ψ r ,v r ,ω r ,t r ], where x r ,y r is the robot coordinate, ψ r is the robot's heading angle, v r =[v x ,v y ] is the robot's moving speed, and v o >>v r ,ω r is the robot's turning angular velocity, t r is the time corresponding to the current state of the robot.

[0012] Furthermore, the specific process of step 2 is as follows: according to the state information, based on the time dimension, the three-dimensional space coordinate system XYZ is expanded, a time coordinate axis T is added on the basis of the three-dimensional coordinate system, and a four-dimensional space-time coordinate system XYZT is established. Each point in the space-time coordinate system corresponds to a specific position in the three-dimensional space and the time at the position; at this time, the coordinates of the obstacle are

[0013] The coordinates of the obstacles within the specified time period can be obtained by fitting the trajectory using polynomials through the acquired state sequence points, and substituting the corresponding coordinate information into the trajectory equation for prediction. At this time, the space-time trajectory of the obstacle within the specified time period can be obtained.

[0014] Furthermore, in step three, the minimum distance between the robot sampling moment and the space-time trajectory corresponding to the high-speed dynamic obstacle in the risk assessment time period is calculated, and the specific steps are:

[0015] The height of the mobile robot is H. High-speed dynamic obstacles will only pose a collision risk to the robot when the height is within the interval [0, H + Δh], where Δh represents the safety height set to increase safety. This height interval is defined as the risk assessment interval, and its corresponding time period [t g ,t h ] is the risk assessment period;

[0016] In the space-time coordinate system, calculate the robot t j The minimum distance between the time and space trajectory corresponding to the high-speed dynamic obstacle in the risk assessment period As the relative distance information between the robot and high-speed dynamic obstacles;

[0017]

[0018] Furthermore, in order to reduce the impact of the unstructured shape of obstacles and mobile robots on the calculation process, a directed bounding box is used to enclose the obstacles and the robot, and the minimum safety distance is defined as follows:

[0019]

[0020] In the formula, l r 、w r 、h r are the length, width and height of the robot envelope, is the length, width and height of the envelope box of the i-th obstacle.

[0021] Further, in the step 4, a reinforcement learning environment (S, A, P, R) for the mobile robot to avoid obstacles in dynamic scenes is constructed;

[0022] S is the state space of the mobile robot,

[0023] A is the action space of the mobile robot, and Satisfy the constraints: For robots in t j The minimum and maximum line speed that can be achieved at any time, For robots in t j The minimum and maximum angular velocity that can be achieved at any time;

[0024] P is the state transfer function of the mobile robot, which is as follows:

[0025]

[0026] R is the reward function, which includes rewards based on obstacle distance and speed Target distance bonus Reward function for goal orientation

[0027] Furthermore, in step 5, a reward based on obstacle speed and distance is designed. Target distance bonus Reward function for goal orientation Guide the mobile robot to successfully avoid obstacles and reach the target point; the designed reward function is as follows:

[0028] Bonuses based on obstacle distance and speed:

[0029]

[0030] in tj The minimum distance of the space-time trajectory corresponding to the i-th high-speed obstacle risk assessment time period from the mobile robot to the i-th high-speed obstacle risk assessment time period, is the minimum safe distance, t j The maximum distance that the mobile robot can reach in the space-time trajectory corresponding to the high-speed obstacle risk assessment period, t j The speed of the i-th obstacle at the moment, is the maximum speed of the obstacle;

[0031] Target distance bonus:

[0032]

[0033] in t j The distance between the moving robot and the target point at any moment, d sg is the distance between the starting point and the target point; the reward function indicates that t j The closer the robot is to the target point at any moment, the higher the reward it gets. When the robot reaches the target point, it will receive an additional higher reward.

[0034] Goal-to-go reward function:

[0035]

[0036] in, t j The orientation angle of the mobile robot at any moment, ψ e is the desired heading angle; the reward function represents, t j The smaller the deviation between the robot's heading angle and the expected heading angle, the greater the reward.

[0037] The reward function for the mobile robot to avoid obstacles is:

[0038]

[0039] Where t0, t n Respectively represent the time when the robot is at the starting point and the target point, λ1, λ2, λ3 are weight coefficients, and satisfy the following constraints:

[0040]

[0041] Another object of the present invention is to provide a mobile robot obstacle avoidance system facing high-speed dynamic obstacles, comprising:

[0042] A state acquisition module is used to obtain the states of obstacles and mobile robots through sensors;

[0043] A coordinate system establishment module is used to expand the three-dimensional space coordinate system based on the time dimension according to the state information to establish a four-dimensional space-time coordinate system;

[0044] A calculation module, used to calculate the minimum distance between the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling moment in the space-time coordinate system;

[0045] The learning environment building module is used to build a reinforcement learning environment for mobile robots to avoid obstacles in dynamic scenes;

[0046] The guidance module is used to design reward functions based on obstacle speed and distance, target distance, and target orientation to guide the mobile robot to successfully avoid obstacles and reach the target point.

[0047] Another object of the present invention is to provide a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the mobile robot obstacle avoidance method facing high-speed dynamic obstacles.

[0048] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the mobile robot obstacle avoidance method facing high-speed dynamic obstacles.

[0049] Another object of the present invention is to provide an information data processing terminal, which is used to implement the mobile robot obstacle avoidance system facing high-speed dynamic obstacles.

[0050] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0051] First, in view of the technical problems existing in the above-mentioned prior art, some creative technical effects are brought about after solving the problems. The specific description is as follows:

[0052] In view of the problems and defects existing in the prior art, the present invention provides a mobile robot obstacle avoidance method for dealing with high-speed dynamic obstacles. The method uses the spatiotemporal trajectory position information of the high-speed dynamic obstacle in the risk assessment time period to replace the position information at discrete sampling moments. When calculating the risk assessment distance information, the minimum value of the spatiotemporal trajectory of the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling moment is calculated, thereby enabling the robot to take evasive action in advance at the sampling moment to avoid the high-speed dynamic obstacle. The present invention first adds a time axis on the basis of a three-dimensional space coordinate system, and establishes a four-dimensional space-time coordinate system. By calculating the minimum distance between the space-time trajectory of the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling moment in the space-time coordinate system, the traditional method overcomes the deficiency that when facing high-speed dynamic obstacles, the position increment of the high-speed dynamic obstacle at adjacent sampling moments is much larger than the position increment of the robot, which causes the calculated risk assessment distance to lead to wrong obstacle avoidance decisions; then, reinforcement learning is used to select the actions taken by the robot, and the robot control instructions are directly obtained to reduce the solution time of the instructions; finally, based on the obstacle distance reward, an additional reward related to the obstacle speed is added, so that the mobile robot can give priority to avoiding obstacles with faster speeds, and stay as far away from the faster obstacles as possible during the avoidance process, thereby improving the safety of the obstacle avoidance process. The advantages and positive effects of the technical solution of the present invention are specifically described as follows:

[0053] (1) The present invention can calculate the minimum distance between the robot and the spatiotemporal trajectory of a high-speed dynamic obstacle at the sampling moment within the risk assessment time period, and can prevent the problem of decision-making failure caused by the large position increment of the high-speed dynamic obstacle at adjacent sampling moments, so that the robot can effectively deal with high-speed dynamic obstacles.

[0054] (2) The process of the present invention is simple, and the planning control instructions of the robot can be directly obtained. In addition, the robot actions are optimized through the reward function, so that the mobile robot can smoothly avoid obstacles and reach the target point while meeting the path optimization.

[0055] (3) The designed reward function related to obstacles can make the robot stay as far away from faster obstacles as possible, thereby improving its own safety. The reward function can also adapt to different obstacle avoidance tasks by adjusting the weights, such as giving priority to avoiding obstacles, giving priority to reaching the target point, etc.

[0056] Second, the technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0057] In view of the problems and defects of the prior art that, when calculating the risk assessment distance, only the distance between the robot and the obstacle at the sampling moment is calculated based on time segment sampling, which makes the robot decision invalid, a mobile robot obstacle avoidance method for dealing with high-speed dynamic obstacles is provided. The method uses the spatiotemporal trajectory position information of the high-speed dynamic obstacle in the risk assessment time period to replace the position information at discrete sampling moments. When calculating the risk assessment distance information, the minimum value of the spatiotemporal trajectory between the robot and the high-speed dynamic obstacle in the risk assessment time period is calculated at the sampling moment, so that the robot can take evasive action in advance at the sampling moment to avoid the high-speed dynamic obstacle.

[0058] The technical solution of the present invention solves a technical problem that people have been eager to solve but have never been successful: the present invention can better deal with high-speed dynamic obstacles. When calculating the risk assessment distance, the present invention uses the minimum distance between the robot and the spatiotemporal trajectory of the high-speed dynamic obstacle in the risk assessment time period at the sampling moment to prevent the decision-making from being invalid due to the large position increment of the high-speed dynamic obstacle at adjacent sampling moments, so that the robot can effectively deal with high-speed dynamic obstacles; and the obstacle-related reward function designed by the present invention can make the robot as far away as possible from faster obstacles, improve its own safety, and have a good response effect to high-speed dynamic obstacles.

[0059] Third, the present invention proposes innovative solutions to multiple technical problems existing in the prior art in the obstacle avoidance process of mobile robots. Traditional obstacle avoidance methods usually rely on simple two-dimensional or three-dimensional coordinate systems. This method has serious limitations when dealing with high-speed dynamic obstacles. It is difficult to accurately predict the movement trajectory of obstacles, resulting in insufficient safety and effectiveness in the obstacle avoidance process. In addition, the prior art usually adopts obstacle avoidance strategies based on fixed rules. These strategies lack flexibility and cannot dynamically adapt to high-speed changes in complex environments, increasing the risk of collision between robots and obstacles.

[0060] In order to solve these problems, the present invention proposes an obstacle avoidance method based on a four-dimensional space-time coordinate system, which introduces the time dimension into the three-dimensional space coordinate system and establishes a four-dimensional space-time model. By calculating the minimum distance between the robot and high-speed dynamic obstacles in the space-time coordinate system, the system can more accurately assess the risk of collision and achieve more efficient obstacle avoidance by predicting the future trajectory of the obstacle. This method significantly improves the robot's obstacle avoidance ability in a high-speed dynamic environment and reduces the probability of accidental collisions.

[0061] In addition, the present invention uses reinforcement learning technology to build an intelligent obstacle avoidance environment for mobile robots in dynamic scenes. By designing a reward function based on obstacle speed, distance, target distance and target orientation, the robot is guided to optimize the obstacle avoidance path, enabling it to autonomously learn and adapt in a complex dynamic environment. Compared with the traditional rule-driven method, the reinforcement learning model can continuously improve the robot's behavior, improve the efficiency and success rate of obstacle avoidance, and achieve more intelligent and adaptive obstacle avoidance decisions.

[0062] The present invention has made significant technological progress in the field of mobile robot obstacle avoidance technology. By introducing a four-dimensional space-time coordinate system and a reinforcement learning algorithm, the present invention overcomes the limitations of the prior art, enabling the robot to have stronger perception, prediction and decision-making capabilities when facing high-speed dynamic obstacles, significantly improving the safety, flexibility and intelligence level of the system, and promoting the further development of mobile robot obstacle avoidance technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a flow chart of a mobile robot obstacle avoidance method for high-speed dynamic obstacles provided by an embodiment of the present invention.

[0064] Figure 2 The present invention provides a block diagram of a mobile robot obstacle avoidance system for high-speed dynamic obstacles.

[0065] Figure 3 It is a flow chart of an algorithm provided by an embodiment of the present invention.

[0066] Figure 4 It is a schematic diagram of the process of establishing a four-dimensional space-time coordinate system provided by an embodiment of the present invention.

[0067] Figure 5 It is a calculation diagram of the relative position information of the robot and the obstacle provided by an embodiment of the present invention.

[0068] Figure 6 It is a schematic diagram of a scene simulation result provided by an embodiment of the present invention.

[0069] Figure 7 4 is a diagram of robot speed and angular velocity simulation results provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0071] The following are two industrial application examples of mobile robot obstacle avoidance methods for high-speed dynamic obstacles:

[0072] Example 1: Autonomous Mobile Robot Obstacle Avoidance in an Automated Logistics Warehouse

[0073] In modern automated logistics warehouses, autonomous mobile robots (AGVs) are widely used for cargo handling and sorting. However, the warehouse environment is complex and there are a large number of dynamic obstacles (such as other AGVs, workers, stacked goods, etc.). These obstacles move fast and change direction frequently, so an efficient obstacle avoidance method is needed to ensure that AGVs can complete their tasks safely and efficiently.

[0074] The obstacle avoidance method for high-speed dynamic obstacles of the present invention has been applied in automated logistics warehouses. AGV obtains obstacle information (including position, speed, etc.) in the surrounding environment in real time through installed laser radar, camera and other sensors. These data are used to construct a four-dimensional space-time coordinate system, analyze the movement trajectory of obstacles and predict their future positions.

[0075] On this basis, the AGV calculates the minimum safe distance between itself and the obstacle, and optimizes the obstacle avoidance path through the reinforcement learning model. The model takes into account the speed, distance and location of the obstacle, and guides the AGV to avoid obstacles and eventually reach the target point by designing a reasonable reward function. This obstacle avoidance method effectively reduces the risk of AGV colliding with other moving obstacles in the warehouse, and significantly improves the safety and efficiency of logistics operations.

[0076] Example 2: Obstacle avoidance for autonomous vehicles in intelligent transportation systems

[0077] In intelligent transportation systems, autonomous vehicles face the challenge of high-speed dynamic obstacles, such as other vehicles, pedestrians, bicycles, etc. The movement of these obstacles on the road is highly uncertain, requiring the autonomous driving system to have the ability to respond quickly and accurately avoid obstacles to ensure the safety of passengers and road users.

[0078] The obstacle avoidance method of the present invention is applied to the control system of an autonomous driving vehicle. Through the radar, laser scanner and camera equipped on the vehicle, the system collects dynamic information of the surrounding environment in real time, identifies and tracks obstacles moving at high speed. These data are used to construct a four-dimensional space-time coordinate system to predict the future trajectory of the obstacle.

[0079] The autonomous driving system uses a reinforcement learning model to calculate the minimum distance between the vehicle and obstacles in the space-time coordinate system and generate the optimal obstacle avoidance strategy based on the actual road conditions. This strategy not only considers the movement speed and distance of the obstacle, but also includes the optimization of the vehicle's orientation and target path. Under the guidance of the reward function, the vehicle can safely avoid other vehicles and pedestrians in a complex traffic environment and successfully reach its destination.

[0080] This method improves the safety and response speed of autonomous vehicles in complex road environments, especially has significant advantages when dealing with high-speed dynamic obstacles, and promotes the development of intelligent transportation systems.

[0081] like Figure 1 As shown, a mobile robot obstacle avoidance method for high-speed dynamic obstacles provided by an embodiment of the present invention comprises the following steps:

[0082] S101, obtaining the status of obstacles and the mobile robot through sensors;

[0083] S102, according to the state information, the three-dimensional space coordinate system is expanded based on the time dimension to establish a four-dimensional space-time coordinate system;

[0084] S103, calculating, in the space-time coordinate system, the minimum distance between the robot and the high-speed dynamic obstacle in the space-time trajectory corresponding to the risk assessment time period at the sampling moment;

[0085] S104, constructing a reinforcement learning environment for mobile robots to avoid obstacles in dynamic scenes;

[0086] S105, design a reward function based on obstacle speed and distance, target distance reward, and target orientation to guide the mobile robot to successfully avoid obstacles and reach the target point.

[0087] In S101 provided in the embodiment of the present invention, the high-speed dynamic obstacle state obtained by the robot sensor is in, is the coordinate of the ith obstacle in the three-dimensional space coordinate system, is the moving speed of the ith obstacle in the three-dimensional space coordinate system, is the moment corresponding to the current state of the ith obstacle; the state of the robot is S r =[x r ,y r ,ψ r ,v r ,ω r ,t r ], where x r ,y r is the robot coordinate, ψ r is the robot's heading angle, v r =[v x ,v y ] is the robot's moving speed, and v o >>v r ,ω r is the robot's turning angular velocity, t r is the time corresponding to the current state of the robot.

[0088] The specific process of S102 provided in the embodiment of the present invention is as follows: according to the state information, based on the time dimension, the three-dimensional space coordinate system XYZ is expanded, a time coordinate axis T is added on the basis of the three-dimensional coordinate system, and a four-dimensional space-time coordinate system XYZT is established. Each point in the space-time coordinate system corresponds to a specific position in the three-dimensional space and the time at the position; at this time, the coordinates of the obstacle are

[0089] The coordinates of the obstacles within the specified time period can be obtained by fitting the trajectory using polynomials through the acquired state sequence points, and substituting the corresponding coordinate information into the trajectory equation for prediction. At this time, the space-time trajectory of the obstacle within the specified time period can be obtained.

[0090] In S103 provided in the embodiment of the present invention, the minimum distance between the robot sampling moment and the space-time trajectory corresponding to the high-speed dynamic obstacle in the risk assessment time period is calculated, and the specific S10 is:

[0091] The height of the mobile robot is H. High-speed dynamic obstacles will only pose a collision risk to the robot when the height is within the interval [0, H + Δh], where Δh represents the safety height set to increase safety. This height interval is defined as the risk assessment interval, and its corresponding time period [t g ,t h ] is the risk assessment period;

[0092] In the space-time coordinate system, calculate the robot t j The minimum distance between the time and space trajectory corresponding to the high-speed dynamic obstacle in the risk assessment period As the relative distance information between the robot and high-speed dynamic obstacles;

[0093]

[0094] Furthermore, in order to reduce the impact of the unstructured shape of obstacles and mobile robots on the calculation process, a directed bounding box is used to enclose the obstacles and the robot, and the minimum safety distance is defined as follows:

[0095]

[0096] In the formula, l r 、w r 、h r are the length, width and height of the robot envelope, is the length, width and height of the envelope box of the i-th obstacle.

[0097] In S104 provided in the embodiment of the present invention, a reinforcement learning environment (S, A, P, R) for the mobile robot to avoid obstacles in a dynamic scene is constructed;

[0098] S is the state space of the mobile robot,

[0099] A is the action space of the mobile robot, and Satisfy the constraints: For robots in t j The minimum and maximum line speed that can be achieved at any time, For robots in t j The minimum and maximum angular velocity that can be achieved at any time;

[0100] P is the state transfer function of the mobile robot, which is as follows:

[0101]

[0102] R is the reward function, which includes rewards based on obstacle distance and speed Target distance bonus Reward function for goal orientation

[0103] In S105 provided in the embodiment of the present invention, a reward based on the speed and distance of the obstacle is designed. Target distance bonus Reward function for goal orientation Guide the mobile robot to successfully avoid obstacles and reach the target point; the designed reward function is as follows:

[0104] Bonuses based on obstacle distance and speed:

[0105]

[0106] in t j The minimum distance of the space-time trajectory corresponding to the i-th high-speed obstacle risk assessment time period from the mobile robot to the i-th high-speed obstacle risk assessment time period, is the minimum safe distance, t j The maximum distance that the mobile robot can reach in the space-time trajectory corresponding to the high-speed obstacle risk assessment period, t j The speed of the i-th obstacle at the moment, is the maximum speed of the obstacle;

[0107] Target distance bonus:

[0108]

[0109] in t jThe distance between the moving robot and the target point at any moment, d sg is the distance between the starting point and the target point; the reward function indicates that t j The closer the robot is to the target point at any moment, the higher the reward it gets. When the robot reaches the target point, it will receive an additional higher reward.

[0110] Goal-to-go reward function:

[0111]

[0112] in, t j The orientation angle of the mobile robot at any moment, ψ e is the desired heading angle; the reward function represents, t j The smaller the deviation between the robot's heading angle and the expected heading angle, the greater the reward.

[0113] The reward function for the mobile robot to avoid obstacles is:

[0114]

[0115] Where t0, t n Respectively represent the time when the robot is at the starting point and the target point, λ1, λ2, λ3 are weight coefficients, and satisfy the following constraints:

[0116]

[0117] like Figure 2 As shown, a mobile robot obstacle avoidance system for high-speed dynamic obstacles provided by an embodiment of the present invention includes:

[0118] A state acquisition module is used to obtain the states of obstacles and mobile robots through sensors;

[0119] A coordinate system establishment module is used to expand the three-dimensional space coordinate system based on the time dimension according to the state information to establish a four-dimensional space-time coordinate system;

[0120] A calculation module, used to calculate the minimum distance between the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling moment in the space-time coordinate system;

[0121] The learning environment building module is used to build a reinforcement learning environment for mobile robots to avoid obstacles in dynamic scenes;

[0122] The guidance module is used to design reward functions based on obstacle speed and distance, target distance, and target orientation to guide the mobile robot to successfully avoid obstacles and reach the target point.

[0123] Another object of the present invention is to provide a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the mobile robot obstacle avoidance method facing high-speed dynamic obstacles.

[0124] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the mobile robot obstacle avoidance method facing high-speed dynamic obstacles.

[0125] Another object of the present invention is to provide an information data processing terminal, which is used to implement the mobile robot obstacle avoidance system facing high-speed dynamic obstacles.

[0126] This embodiment uses a wheeled robot as the control object. The wheeled robot is equipped with a sensor that can obtain obstacle information, and the robot's own state can be obtained by the robot's own odometer. A scene with dynamic and static obstacles on the ground and high-speed dynamic obstacles in the air is designed, where the speed of the high-speed dynamic obstacle is 300m / s, the maximum speed of the robot is 15m / s, and the starting point is set to (0,0), the target point is (25,10), the unit is m, and the robot's heading angle is -90°.

[0127] like Figure 3 The flowchart of the algorithm of the present invention is shown. The specific real-time steps of a mobile robot obstacle avoidance method for high-speed dynamic obstacles provided by an embodiment of the present invention are as follows:

[0128] Step 1: Get obstacles and robot status.

[0129] The obstacle status obtained by the robot sensor is in, is the coordinate of the ith obstacle in the three-dimensional space coordinate system, is the moving speed of the ith obstacle in the three-dimensional space coordinate system, is the time corresponding to the current state of the i-th obstacle. The state of the vehicle is S r =[x r ,y r ,ψ r ,v r ,ω r ,t r ], where x r ,y r is the robot coordinate, ψ r is the robot's heading angle, v r =[v x ,v y] is the robot movement speed, ω r is the robot's turning angular velocity, t r is the time corresponding to the current state of the robot.

[0130] Step 2: Establish a four-dimensional space-time coordinate system.

[0131] like Figure 4 As shown in the figure, based on the state information and the time dimension, the three-dimensional space coordinate system XYZ is expanded, and a time coordinate axis T is added on the basis of the three-dimensional coordinate system to establish the four-dimensional space-time coordinate system XYZT. Each point in the space-time coordinate system corresponds to a specific position in the three-dimensional space and the time at that position. At this time, the coordinates of the obstacle are

[0132] Furthermore, the coordinates of the obstacles within the specified time period can be obtained by fitting the trajectory using polynomials through the acquired state sequence points, and substituting the corresponding coordinate information into the trajectory equation for prediction. At this time, the space-time trajectory of the obstacles within the specified time period can be obtained.

[0133] Step three, in the space-time coordinate system, calculate the minimum space-time trajectory of the robot at the sampling moment and the high-speed dynamic obstacle in the risk assessment time period.

[0134] The height of the mobile robot is H. High-speed dynamic obstacles will only pose a collision risk to the robot when the height is within the interval [0, H + Δh], where Δh represents the safety height set to increase safety. This height interval is defined as the risk assessment interval, and its corresponding time period [t g ,t h ] is the risk assessment time period.

[0135] like Figure 5 As shown, in the space-time coordinate system, the robot t is calculated j The minimum distance between the time and space trajectory corresponding to the high-speed dynamic obstacle in the risk assessment period As the distance information between the robot and high-speed dynamic obstacles.

[0136]

[0137] Furthermore, in order to reduce the impact of the unstructured shape of obstacles and mobile robots on the calculation process, a directed bounding box is used to enclose the obstacles and the robot, and the minimum safety distance is defined as follows:

[0138]

[0139] In the formula, l r 、w r 、h rare the length, width and height of the robot envelope, is the length, width and height of the envelope box of the i-th obstacle.

[0140] Step 4: Build a reinforcement learning environment.

[0141] Construct a reinforcement learning environment (S, A, P, R) for mobile robots to avoid obstacles in dynamic scenes.

[0142] S is the state space of the mobile robot,

[0143] A is the action space of the mobile robot, and Satisfy the constraints: For robots in t j The minimum and maximum line speed that can be achieved at any time, The minimum and maximum angular velocities that can be achieved;

[0144] P is the state transfer function of the mobile robot, which is as follows:

[0145]

[0146] R is the reward function, which includes rewards based on obstacle distance and speed, target distance rewards, and target orientation reward functions.

[0147] Step 5: Calculate the reward function and set different weights to guide the robot to avoid obstacles and reach the target point safely according to the main purpose of obstacle avoidance.

[0148] Bonuses based on obstacle distance and speed:

[0149]

[0150] in t j The distance from the mobile robot to the i-th obstacle at any moment, is the minimum safe distance, t j The maximum distance that the mobile robot can reach from the obstacle at any time. t j The speed of the i-th obstacle at the moment, is the maximum speed of the obstacle. The reward function gives higher rewards when the robot is farther away from the faster obstacle. That is, when avoiding obstacles, the robot will take the action of avoiding faster obstacles first. And the reward function will make the robot take the fastest action possible within the constraints of speed and angular velocity to make the distance between itself and the obstacle larger, further improving safety. When the distance between the robot and the obstacle is less than the safety value, it will be punished.

[0151] Target distance bonus:

[0152]

[0153] in t j The distance between the moving robot and the target point at any moment, d sg is the distance between the starting point and the target point. The reward function indicates that t j The closer the robot is to the target point at any moment, the higher the reward it gets. When the robot reaches the target point, it will receive an additional higher reward.

[0154] Goal-to-go reward function:

[0155]

[0156] in t j The orientation angle of the mobile robot at any moment, ψ e is the desired heading angle. This reward function indicates that t j The smaller the difference between the robot's orientation angle and the expected orientation angle, the greater the reward.

[0157] The reward function for the mobile robot to avoid obstacles is:

[0158]

[0159] Where t0, t n Respectively represent the time when the robot is at the starting point and the target point, λ1, λ2, λ3 are weight coefficients, and satisfy the following constraints:

[0160]

[0161] In the process of obstacle avoidance, the robot's primary goal is to avoid obstacles and ensure its own safety, and the second is to reach the target point. The heading angle constraint can be added as needed. The weight coefficient can be set according to the priority of the obstacle avoidance task, such as giving priority to obstacle avoidance or reaching the target point.

[0162] In this example, the robot's obstacle avoidance task is to prioritize staying away from obstacles. Under the constraint conditions, the weight coefficients are set as: λ1=0.7, λ1=0.2, λ1=0.1, and finally the following is obtained: Figure 6 The simulation results are shown in Figure 1. The black track is the robot's driving track, and the red track is the obstacle's driving track. The results show that the robot successfully avoids high-speed dynamic obstacles and reaches the target point. Its heading angle at the target point is -86.4°, which is 3.6° away from the expected value. At the same time, in the process of heading to the target point, when facing high-speed dynamic obstacles, the robot's trajectory shows a trend of staying as far away as possible, which is in line with the obstacle avoidance task of keeping away from obstacles first. The results of the robot's speed and angular velocity changes during the obstacle avoidance process are shown in Figure 1. Figure 7 As shown, according to the results, it can be seen that the robot takes acceleration action when encountering high-speed dynamic obstacles.

[0163] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0164] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A mobile robot obstacle avoidance method facing high-speed dynamic obstacles, characterized in that: The following steps are involved: Step 1: Obtain the status of obstacles and mobile robots through sensors; Step 2: According to the state information, the three-dimensional space coordinate system is expanded based on the time dimension to establish a four-dimensional space-time coordinate system; Step 3: In the space-time coordinate system, calculate the minimum distance between the robot and the high-speed dynamic obstacle at the sampling moment in the space-time trajectory corresponding to the risk assessment time period; Step 4: Build a reinforcement learning environment for mobile robots to avoid obstacles in dynamic scenes; Step 5: Design a reward function based on obstacle speed and distance, target distance, and target orientation to guide the mobile robot to successfully avoid obstacles and reach the target point.

2. The mobile robot obstacle avoidance method facing high-speed dynamic obstacles as claimed in claim 1, characterized in that: In step 1, the high-speed dynamic obstacle state obtained by the robot sensor is in, is the coordinate of the ith obstacle in the three-dimensional space coordinate system, is the moving speed of the ith obstacle in the three-dimensional space coordinate system, is the moment corresponding to the current state of the ith obstacle; the state of the robot is S r =[x r ,y r ,ψ r ,v r ,ω r ,t r ], where x r ,y r is the robot coordinate, ψ r is the robot's heading angle, v r =[v x ,v y ] is the robot's moving speed, and v o >>v r ,ω r is the robot's turning angular velocity, t r is the time corresponding to the current state of the robot.

3. The mobile robot obstacle avoidance method facing high-speed dynamic obstacles as claimed in claim 1, characterized in that: The specific process of step 2 is as follows: according to the state information, based on the time dimension, the three-dimensional space coordinate system XYZ is expanded, a time coordinate axis T is added on the basis of the three-dimensional coordinate system, and a four-dimensional space-time coordinate system XYZT is established. Each point in the space-time coordinate system corresponds to a specific position in the three-dimensional space and the time at that position; at this time, the coordinates of the obstacle are The coordinates of the obstacles within the specified time period can be obtained by fitting the trajectory using polynomials through the acquired state sequence points, and substituting the corresponding coordinate information into the trajectory equation for prediction. At this time, the space-time trajectory of the obstacle within the specified time period can be obtained.

4. The mobile robot obstacle avoidance method facing high-speed dynamic obstacles as claimed in claim 1, characterized in that: In step 3, the minimum distance between the robot sampling moment and the space-time trajectory corresponding to the high-speed dynamic obstacle in the risk assessment time period is calculated, and the specific steps are: The height of the mobile robot is H. High-speed dynamic obstacles will only pose a collision risk to the robot when the height is within the interval [0, H + Δh], where Δh represents the safety height set to increase safety. This height interval is defined as the risk assessment interval, and its corresponding time period [t g ,t h ] is the risk assessment period; In the space-time coordinate system, calculate the robot t j The minimum distance between the time and space trajectory corresponding to the high-speed dynamic obstacle in the risk assessment period As the relative distance information between the robot and high-speed dynamic obstacles; Define the minimum safe distance as follows: In the formula, l r 、w r 、h r are the length, width and height of the robot envelope, is the length, width and height of the envelope box of the i-th obstacle.

5. The mobile robot obstacle avoidance method facing high-speed dynamic obstacles as claimed in claim 1, characterized in that: In the step 4, a reinforcement learning environment (S, A, P, R) for the mobile robot to avoid obstacles in dynamic scenes is constructed; S is the state space of the mobile robot, A is the action space of the mobile robot, and Satisfy the constraints: For robots in t j The minimum and maximum line speed that can be achieved at any time, For robots in t j The minimum and maximum angular velocity that can be achieved at any time; P is the state transfer function of the mobile robot, which is as follows: R is the reward function, which includes rewards based on obstacle distance and speed Target distance bonus Reward function for goal orientation 6. The mobile robot obstacle avoidance method facing high-speed dynamic obstacles as claimed in claim 1, characterized in that: In step 5, design rewards based on obstacle speed and distance Target distance bonus Reward function for goal orientation Guide the mobile robot to successfully avoid obstacles and reach the target point; the designed reward function is as follows: Bonuses based on obstacle distance and speed: in t j The minimum distance of the space-time trajectory corresponding to the i-th high-speed obstacle risk assessment time period from the mobile robot to the i-th high-speed obstacle risk assessment time period, is the minimum safe distance, t j The maximum distance that the mobile robot can reach in the space-time trajectory corresponding to the high-speed obstacle risk assessment period, t j The speed of the i-th obstacle at the moment, is the maximum speed of the obstacle; Target distance bonus: in t j The distance between the moving robot and the target point at any moment, d sg is the distance between the starting point and the target point; the reward function indicates that t j The closer the robot is to the target point at any moment, the higher the reward it gets. When the robot reaches the target point, it will receive an additional higher reward. Goal-to-go reward function: in, t j The orientation angle of the mobile robot at any moment, ψ e is the desired heading angle; the reward function represents, t j The smaller the deviation between the robot's heading angle and the expected heading angle, the greater the reward. The reward function for the mobile robot to avoid obstacles is: Where t0, t n Respectively represent the time when the robot is at the starting point and the target point, λ1, λ2, λ3 are weight coefficients, and satisfy the following constraints:

7. A mobile robot obstacle avoidance system facing high-speed dynamic obstacles implementing the mobile robot obstacle avoidance method facing high-speed dynamic obstacles as claimed in any one of claims 1 to 6, characterized in that: The mobile robot obstacle avoidance system for high-speed dynamic obstacles comprises: A state acquisition module is used to obtain the states of obstacles and mobile robots through sensors; A coordinate system establishment module is used to expand the three-dimensional space coordinate system based on the time dimension according to the state information to establish a four-dimensional space-time coordinate system; A calculation module, used to calculate the minimum distance between the robot and the high-speed dynamic obstacle in the risk assessment time period at the sampling moment in the space-time coordinate system; The learning environment building module is used to build a reinforcement learning environment for mobile robots to avoid obstacles in dynamic scenes; The guidance module is used to design reward functions based on obstacle speed and distance, target distance, and target orientation to guide the mobile robot to successfully avoid obstacles and reach the target point.

8. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the mobile robot obstacle avoidance method facing high-speed dynamic obstacles as described in any one of claims 1-6.

9. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the mobile robot obstacle avoidance method facing high-speed dynamic obstacles as described in any one of claims 1 to 6.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the mobile robot obstacle avoidance system for high-speed dynamic obstacles as described in claim 7.

Citation Information

Cited By

  • Obstacle avoidance method and system for automatically guiding logistics robot

    CN119759037A