A robot obstacle avoidance method suitable for complex environments

Abstract

The present invention discloses a robot obstacle avoidance method suitable for complex environments, belonging to the field of robot obstacle avoidance technology, and its technical points are: comprising the following steps: Step 1: Scanning and mapping the operating environment, decomposing the complex graphics, and using multiple super ellipse functions to cover each part; Step 2: Designing a CBF controller, wherein the CBF controller designs a control strategy according to the distance and direction between the robot and the coverage area, and has the advantages of more accurately describing and covering complex graphics by using multiple super ellipse functions, thereby improving the accuracy and efficiency of obstacle avoidance. Secondly, the parameters of the super ellipse function can be adjusted according to the shape and size of each part, thereby improving the flexibility and adaptability of the method. In addition, by designing a corresponding control strategy according to the distance and direction between the intelligent body and the coverage area, the advantage of safe movement of the intelligent body in a complex environment can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of robot obstacle avoidance, and in particular to a robot obstacle avoidance method suitable for use in complex environments. Background Art

[0002] Intelligent agent obstacle avoidance algorithm is an important research direction in the field of artificial intelligence and robotics. With the continuous development of technologies such as autonomous mobile robots, self-driving cars, and drones, the need for intelligent agents to safely navigate and avoid obstacles in complex environments is becoming increasingly urgent. Traditional robot navigation and path planning methods usually use static maps or environmental models to pre-plan the robot's movement path. However, even in static environments, these methods face the limitation of not being able to effectively handle complex obstacle shapes. Therefore, the research on intelligent agent obstacle avoidance algorithms aims to develop algorithms that can adapt to complex obstacle environments to ensure that intelligent agents can safely avoid obstacles and reach the target location.

[0003] In past studies, many scholars have proposed various obstacle avoidance algorithms to solve the problem of intelligent agents navigating in complex environments. These algorithms include but are not limited to the traditional potential field method, A* algorithm, Dijkstra algorithm, etc. Although these algorithms have shown good results in some scenarios, they also have some limitations.

[0004] Potential field method: Potential field method is a common obstacle avoidance method. It introduces a virtual potential field in the environment to simulate the impact of obstacles on the agent, thereby guiding the agent to move to the target location. However, the potential field method may have the problem of local minimum, causing the agent to fall into an infinite loop or fail to reach the target.

[0005] A* algorithm: A* algorithm is a heuristic search algorithm that determines the optimal path by estimating the cost from each node to the target. Although A* algorithm performs well in static environments, it has limited processing capabilities for dynamic environments and cannot adapt to environmental changes in a timely manner.

[0006] Dijkstra algorithm: Dijkstra algorithm is a commonly used graph search algorithm for finding the shortest path from the starting point to the end point. However, the computational complexity of Dijkstra algorithm is high, especially in large-scale environments, the running time of the algorithm may be very long.

[0007] In summary, although traditional obstacle avoidance algorithms show good results in some scenarios, they still have some limitations, such as weak processing capabilities for complex obstacle shapes and insufficient adaptability to dynamic environments. Therefore, it is necessary to propose a novel obstacle avoidance method to overcome these limitations and provide a more reliable and efficient solution for the navigation of intelligent agents in complex environments.

[0008] In recent years, machine learning and deep learning technologies have been widely used in intelligent agent obstacle avoidance algorithms. By training neural network models, intelligent agents can learn environmental features and the movement patterns of obstacles from sensor data, so as to avoid obstacles more effectively. Although some progress has been made in intelligent agent obstacle avoidance algorithms, there are still some challenges and limitations. First, with the increase of environmental complexity, traditional path planning algorithms often face problems such as dimensionality curse and search space explosion, resulting in a sharp increase in computational complexity. Second, the accuracy and reliability of sensors may be limited by environmental conditions, such as bad weather and poor lighting conditions. In addition, there are still certain challenges for intelligent agents when dealing with complex situations such as dynamic obstacles and multi-agent collaborative obstacle avoidance. In summary, the research on intelligent agent obstacle avoidance algorithms has important theoretical and practical significance, but further in-depth research and exploration are still needed to cope with challenges in complex environments and provide more reliable and efficient solutions for the safe movement and navigation of intelligent agents in the real world. Summary of the invention

[0009] In view of the deficiencies in the prior art, an embodiment of the present invention aims to provide a robot obstacle avoidance method suitable for use in complex environments, so as to solve the problems in the above-mentioned background technology.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] A robot obstacle avoidance method suitable for use in complex environments comprises the following steps:

[0012] Step 1: Scan and map the operating environment, decompose the complex graphics, and use multiple super ellipse functions to cover each part;

[0013] Step 2: Design a CBF controller, wherein the CBF controller designs a control strategy according to the distance and direction between the robot and the coverage area.

[0014] As a further solution of the present invention, the complex graphics in step 1 include T-shaped, L-shaped, and cross-shaped obstacles, and two super ellipse functions are used to cover the T-shaped obstacle, one super ellipse will cover the horizontal part, and the other super ellipse will cover the vertical part.

[0015] As a further solution of the present invention, each rectangular area of ​​the obstacle is represented by the formula:

[0016] m square ={(x, y)|x min ≤x≤x max ,y min ≤y≤y max},

[0017] The center of the superellipse covering each rectangle (x oi ,y oi ), semi-major axis a oi and semi-minor axis b oi is represented as:

[0018]

[0019] As a further solution of the present invention, the complex figure in step 1 includes a five-pointed star-shaped obstacle, covering a regular five-pointed star-shaped obstacle with the following coordinate information: the center point coordinate is (x c ,y c ), the coordinates of the five external vertices are (x1, y1), (x2, y2), (x3, y3), (x4, y4), (x5, y5), and the side length is s.

[0020] As a further solution of the present invention, the parameters of the three superellipse functions of the five-pointed star-shaped obstacle are adjusted. For the horizontal superellipse, the center coordinate (x oi ,y oi ) and the coordinates of the center point of the five-pointed star (x c ,y c )same.

[0021] Major axis radius a oi It is determined according to the horizontal range of the five-pointed star, which is half the diameter of the circumscribed circle of the five-pointed star.

[0022]

[0023] As a further solution of the present invention, the parameters of the three superellipse functions of the five-pointed star-shaped obstacle are adjusted. For the vertical oblique superellipse, the center coordinate (x oi ,y oi ) Select the midpoints on either side of the star:

[0024]

[0025] The major axis radius and minor axis radius are determined by the shortest distance from the center coordinate to the boundary of the five-pointed star.

[0026]

[0027] As a further solution of the present invention, the complex graph is divided into a safe part S and an unsafe part U, both of which satisfy:

[0028] S∪U=R n ,

[0029] And the security part S is defined as follows:

[0030]

[0031] Among them, (x i ,y i ) is the center of the ith obstacle in the map environment, ri is the farthest distance from the obstacle center to the obstacle, θ is the rotation angle of the superellipse, a and b are the semi-major axis and semi-minor axis of the superellipse, and n is the index that controls the shape of the superellipse (when n = 1, the safe area is an ellipse). Therefore, the safe area part S can be expressed as:

[0032]

[0033] As a further solution of the present invention, the CBF controller includes a control barrier function (Relaxed CBF), and for a system and a known safety region S, the Relaxed CBF has a continuously differentiable function H:

[0034] X→R n ,

[0035] If the following three conditions are met, it can be called a relaxed control barrier function.

[0036] For all x∈S, H(x)≥0, for any M∈R, the set {x|H(x)≤M} is closed, and for any continuous mapping u h : R n ×R→R m , there exist continuous non-negative functions α, β: R→R≥0, such that the following equation holds:

[0037]

[0038] In summary, compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0039] It has strong robustness and flexibility, and can ensure that the robot can achieve path tracking while avoiding obstacles. By using super elliptical functions to cover complex graphics and combining CBF controllers, it can effectively cope with obstacle avoidance challenges in a variety of complex environments.

[0040] In order to more clearly illustrate the structural features and effects of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a structural schematic diagram of an embodiment of the invention.

[0042] Figure 2 It is a super ellipse covering T-shaped and L-shaped obstacles in the embodiment of the invention.

[0043] Figure 3 It is a superellipse covering a five-pointed star area in the embodiment of the invention. DETAILED DESCRIPTION

[0044] 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 accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0045] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0046] In one embodiment, a robot obstacle avoidance method suitable for use in complex environments, see Figure 1 to Figure 3 , including the following steps:

[0047] Step 1: Scan and map the operating environment, decompose the complex graphics, and use multiple super ellipse functions to cover each part;

[0048] Step 2: Design a CBF controller, wherein the CBF controller designs a control strategy according to the distance and direction between the robot and the coverage area.

[0049] For further information, see Figure 1 to Figure 3 In the step 1, the complex graphics include T-shaped, L-shaped, and cross-shaped obstacles. Two super ellipse functions are used to cover the T-shaped obstacle, one super ellipse will cover the horizontal part, and the other super ellipse will cover the vertical part.

[0050] For further information, see Figure 1 to Figure 3 , each rectangular area of ​​the obstacle uses the formula:

[0051] m square ={(x, y)|x min ≤x≤x max ,y min ≤y≤y max},

[0052] The center of the superellipse covering each rectangle (x oi ,y oi ), semi-major axis a oi and semi-minor axis b oi is represented as:

[0053]

[0054] For further information, see Figure 1 to Figure 3 The complex figure in step 1 includes a five-pointed star obstacle, covering a regular five-pointed star obstacle with the following coordinate information: the center point coordinate is (x c ,yc ), the coordinates of the five external vertices are (x1, y1), (x2, y2), (x3, y3), (x4, y4), (x5, y5), and the side length is s.

[0055] For further information, see Figure 1 to Figure 3 , the parameters of the three superellipse functions of the five-pointed star-shaped obstacle are adjusted. For the horizontal superellipse, the center coordinate (x oi ,y oi ) and the coordinates of the center point of the five-pointed star (x c ,y c )same.

[0056] Major axis radius a o i is determined according to the horizontal range of the five-pointed star, which is half the diameter of the circumscribed circle of the five-pointed star.

[0057]

[0058] For further information, see Figure 1 to Figure 3 , the parameters of the three superellipse functions of the five-pointed star-shaped obstacle are adjusted. For the vertical superellipse, the center coordinate (x oi ,y oi ) Select the midpoints on either side of the star:

[0059]

[0060] The major axis radius and minor axis radius are determined by the shortest distance from the center coordinate to the boundary of the five-pointed star.

[0061]

[0062] For further information, see Figure 1 to Figure 3 , the complex graph is divided into a safe part S and an unsafe part U, both of which satisfy:

[0063] S∪U=R n

[0064] And the security part S is defined as follows:

[0065]

[0066] Among them, (x i ,y i ) is the center of the ith obstacle in the map environment, ri is the farthest distance from the obstacle center to the obstacle, θ is the rotation angle of the superellipse, a and b are the semi-major axis and semi-minor axis of the superellipse, and n is the index that controls the shape of the superellipse (when n = 1, the safe area is an ellipse). Therefore, the safe area part S can be expressed as:

[0067]

[0068] For further information, see Figure 1 to Figure 3 , the CBF controller includes a control barrier function (Relaxed CBF), and the Relaxed CBF has a continuous differentiable function H for a system and a known safety region S:

[0069] X→R n ,

[0070] If the following three conditions are met, it can be called a relaxed control barrier function.

[0071] For all x∈S, H(x)≥0. For any M∈R, the set {x|H(x)≤M} is closed, and for any continuous mapping u h : R n ×R→R m , there exist continuous non-negative functions α, β: R→R≥0, such that the following equation holds:

[0072]

[0073] R represents the real number domain, which is the conventional one-dimensional real number space; R n represents a real number space of dimension n; M∈R represents an arbitrary real number; H(x): X→R n represents a designed control obstacle function; t represents the time domain; u∈R2 represents the input of the obstacle avoidance controller; u h ∈R2 represents the input of other controllers; g(x) represents the state transfer function of the smart car,

[0074] In this embodiment, the path planning navigation and path tracking of a single robot are realized. The system includes a robot equipped with a multi-line laser radar to perceive the environment, has communication capabilities to communicate with external devices, and has simple computing capabilities, and can perform path planning and path tracking according to instructions.

[0075] The host computer in the system is responsible for sending path planning instructions to the robot. The robot navigates according to the received path planning instructions and senses the environment through lidar to avoid collisions and optimize the path.

[0076] The robot's action space is confined to a two-dimensional plane bounded space to ensure that the robot can move freely within it. In simulation, boundaries can be set to prevent the robot from crossing the boundary. Before the system is actually applied, it is necessary to complete the scanning and mapping of the operating environment to provide an accurate description of the environment and the information required for robot navigation.

[0077] The obstacle avoidance algorithm consists of two key steps. First, the complex figure is covered with a superellipse obstacle function. In this step, the complex figure is decomposed and multiple superellipse functions are used to cover each part. The second step is the design of the CBF controller. The CBF controller designs the corresponding control strategy based on the distance and direction between the robot and the coverage area to ensure that the robot maintains stability and safety while avoiding obstacles. This algorithm combines the coverage of obstacles, the design of obstacle avoidance paths, and the design of controllers to provide a comprehensive solution for the safe movement of robots in complex environments.

[0078] A method of covering convex obstacles with super elliptical obstacle functions is proposed for different shapes of convex obstacles. Convex obstacles include L-shaped obstacles, T-shaped obstacles, cross-shaped obstacles, Y-shaped obstacles, Z-shaped obstacles, pentagram-shaped obstacles, etc. Specifically, for each convex obstacle, we decompose it into multiple parts and use one or more super elliptical functions to cover each part. These super elliptical functions can accurately describe and cover complex obstacle shapes, providing an important basis for the subsequent obstacle avoidance path design. By adopting the method of super elliptical obstacle function coverage, we can improve the accuracy and efficiency of the robot's obstacle avoidance in complex environments, thereby achieving safer and more reliable navigation.

[0079] For T-shaped, L-shaped, and cross-shaped obstacles, two superellipse functions will be used to cover the T-shaped obstacle. One superellipse will cover the horizontal part, and the other superellipse will cover the vertical part. The following is a description of the specific parameters of the obstacle superellipse coverage:

[0080] Assume that each rectangular area constituting an obstacle is expressed by the following formula:

[0081] m square ={(x, y)|x min ≤x≤x max ,y min ≤y≤y max},

[0082] The center of the superellipse covering each rectangle (x oi ,y oi ), semi-major axis a oi and semi-minor axis b oi is represented as:

[0083]

[0084] For the five-pointed star obstacle, suppose we want to cover a regular five-pointed star obstacle with the following coordinate information: the center point coordinate is (x c ,y c). The coordinates of the five outer vertices are (x1, y1), (x2, y2), (x3, y3), (x4, y4), (x5, y5), and the side length is s. Next, we will cover this five-pointed star-shaped obstacle by adjusting the parameters of the three superellipse functions:

[0085] For the transverse superellipse, the center coordinate (x oi ,y oi ) and the coordinates of the center point of the five-pointed star (x c ,y c )same.

[0086] Major axis radius a oi It is determined according to the horizontal range of the five-pointed star, which is half the diameter of the circumscribed circle of the five-pointed star.

[0087]

[0088] For the vertical superellipse, the center coordinate (x oi ,y oi ) Select the midpoints on both sides of the five-pointed star The major axis radius and the minor axis radius are determined according to the shortest distance from the center coordinate to the boundary of the pentagon.

[0089]

[0090] The entire map is divided into two parts: the safe part S and the unsafe part U, which satisfy S∪U=R n , and the security part S is defined as follows:

[0091]

[0092] Among them, (x i ,y i ) is the center of the i-th obstacle in the map environment, r i is the farthest distance from the obstacle center to the obstacle, θ is the rotation angle of the superellipse, a and b are the semi-major axis and semi-minor axis of the superellipse, and n is the index that controls the shape of the superellipse (when n = 1, the safe area is an ellipse). Therefore, the safe area part S can be expressed as:

[0093]

[0094] In the second step, the present invention designs a CBF-based obstacle avoidance controller to guide the agent to stay away from obstacles when approaching an obstacle area. First, a control obstacle function needs to be designed.

[0095] Relaxed CBF is defined as follows: For a system and a known safe region S, there exists a continuously differentiable function H:X→R n, if the following three conditions are met, it can be called a relaxed control barrier function

[0096] (C1) For all x∈S, H(x)≥0

[0097] (C2) For any M∈R, the set {x|H(x)≤M} is closed

[0098] (C3) For any continuous mapping u h :R n ×R→R m , there exist continuous non-negative functions α, β: R→R≥0, n represents n-dimension, m represents m-dimension, and N is also the same value, which means dimension, so that the following equation holds:

[0099]

[0100] It should be noted that for each Relaxed CBF function, H must be infinite on the boundary of the safe set. i ) i∈N ,when

[0101] The designed superellipse-based barrier function is:

[0102]

[0103] Obviously, for any (x,y)∈S, there is h i ≥0, and then H(x)≥0, satisfying condition (C1);

[0104] For any M ≥ 0, the following set is clearly unbounded:

[0105]

[0106] Therefore, the complement of set Y is c is a closed set, satisfying condition (C2);

[0107] Here is the proof of condition (C3):

[0108]

[0109] Let u=-u h Substituting into the above formula, we get the following inequality:

[0110]

[0111] Among them, α and β are non-negative functions, so they satisfy condition (C3);

[0112] In summary, the CBF function we designed meets the above three conditions.

[0113] In order to shorten the time it takes for the robot to cross the obstacle area, it is recommended to move along the safety boundaries of the multiple super ellipses mentioned above. Assume that the super ellipse E n The boundary equation is f n (x, y) = 0. Then we can design a piecewise function g(x, y) and select the corresponding hyperellipse boundary equation according to the position of the smart car.

[0114] The expression of the piecewise function is as follows:

[0115]

[0116] We use the following hyperelliptic function to characterize the impact characteristics of obstacles, conceived as a dynamic energy field, to help plan obstacle avoidance paths:

[0117]

[0118] Where (x, y) represents the position of the robot, Q (x，y) Indicates the energy detected by the robot at the current position. (x，y) > 0, it means the robot is currently in a safe area. In order to find the direction of the minimum gradient on the superellipse, we select a circle with a radius of r and the current position as the center. Then select a set of discrete points on this circle. The coordinates of these discrete points can be expressed as:

[0119] x i =x+rcos(θ i ),

[0120] y i =y+rsin(θ i ),

[0121] Among them, θ i is the angle of the discrete point on the circle. Substituting the coordinates of the discrete point into the energy equation, we get the function value Q for each point on the circle. (x，y) . Select the point that minimizes the following equation from the discrete points, namely:

[0122]

[0123] point will become the next target point of the robot. The robot moves to the target point along the direction of the minimum gradient, and then repeats the above steps to continuously update the target point. In this way, the intelligent agent moves along the boundaries of multiple super elliptical piecewise functions to avoid obstacles.

[0124] The obstacle avoidance controller designed by the present invention is:

[0125] u=u CBF +u PID

[0126] in:

[0127]

[0128] in:

[0129]

[0130] in is a constant. The following proves that the system is stable under the action of this controller u:

[0131] If P>Q, then the above formula can be written as:

[0132]

[0133] If P<Q, then the above formula can be written as:

[0134]

[0135] So the following inequality holds:

[0136]

[0137] For any t ≥ 0, H(X t ) has an upper bound. If Then H is unbounded, so X(t)∈S. Through the above controller, it can be ensured that the agent will not collide with obstacles.

[0138] When P≤Q, the input u of the CBF controller is 0, so the PID controller u PID The agent can be directly controlled. When the agent approaches the obstacle area, P>Q, PID input u PID will be offset, ensuring the robot remains within the safe zone at all times.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A robot obstacle avoidance method suitable for use in complex environments, characterized in that: The following steps are involved: Step 1: Scan and map the operating environment, decompose the complex graphics, and use multiple super ellipse functions to cover each part; The complex graphics include a five-pointed star obstacle, which covers a regular five-pointed star obstacle with the following coordinate information: the center point coordinates are (x c ,y c ), the coordinates of the five external vertices are (x1, y1), (x2, y2), (x3, y3), (x4, y4), (x5, y5), and the side length is s; The parameters of the three superellipse functions of the five-pointed star-shaped obstacle are adjusted. For the horizontal superellipse, the center coordinate (x oi ,y oi ) and the coordinates of the center point of the five-pointed star (x c ,y c )same, Major axis radius a oi Determined according to the horizontal range of the five-pointed star, which is half the diameter of the circumscribed circle of the five-pointed star; (x oi ,and oi )=(x c ,and c ), The parameters of the three superellipse functions of the five-pointed star-shaped obstacle are adjusted. For the vertical superellipse, the center coordinate (x oi ,y oi ) Select the midpoints on either side of the star: The major axis radius and minor axis radius are determined by the shortest distance from the center coordinate to the boundary of the five-pointed star. The complex graph is divided into a safe part S and an unsafe part U, both of which satisfy: S∪U=R n , R n represents the real number space of dimension n; And the security part S is defined as follows: Among them, (x i ,y i ) is the center of the ith obstacle in the map environment, ri is the farthest distance from the obstacle center to the obstacle, θ is the rotation angle of the superellipse, a and b are the semi-major axis and semi-minor axis of the superellipse, and n is the index that controls the shape of the superellipse. When n = 1, the safe area is an ellipse. Therefore, the safe area part S can be expressed as: Step 2: Design a CBF controller, wherein the CBF controller designs a control strategy according to the distance and direction between the robot and the coverage area.

2. The robot obstacle avoidance method suitable for complex environments according to claim 1, characterized in that: The complex graphics in step 1 include T-shaped, L-shaped, and cross-shaped obstacles. Two super ellipse functions are used to cover the T-shaped obstacle, one super ellipse will cover the horizontal part, and the other super ellipse will cover the vertical part.

3. The robot obstacle avoidance method suitable for complex environments according to claim 2, characterized in that: Each rectangular area of ​​the obstacle uses the formula: m square ={(x,y)|x min ≤x≤x max ,y min ≤y≤y max }, The center of the superellipse covering each rectangle (x oi ,y oi ), semi-major axis a oi and semi-minor axis b oi is represented as:

4. The robot obstacle avoidance method suitable for complex environments according to claim 1, characterized in that: The CBF controller includes a control barrier function Relaxed CBF, wherein for a system and a known safety region S, there exists a continuously differentiable function H: X→R n , If the following three conditions are met, it can be called a relaxed control barrier function, where R represents the real number domain, that is, the conventional one-dimensional real number space; R n represents a real number space of dimension n; M∈R represents an arbitrary real number; H(x): X→R n represents a designed control obstacle function; t represents the time domain; u∈R2 represents the input of the obstacle avoidance controller; u h ∈R2 represents the input of other controllers; g(x) represents the state transition function of the smart car. For all x∈S, H(x)≥0. For any M∈R, the set {x|H(x)≤M} is closed. For any continuous mapping u h : R n ×R→R m , m represents m-dimension, there exists a continuous non-negative function α, β: R→R≥0, so that the following equation holds:

Images