Mobile robot dynamic obstacle avoidance method based on multi-feature repulsive potential field

Abstract

The application discloses a mobile robot dynamic obstacle avoidance method based on a multi-feature repulsive potential field, wherein a laser radar, a positioning system, a three-dimensional laser sensor and a four-line laser sensor are arranged on the robot; the laser radar, the positioning system, the three-dimensional laser sensor and the four-line laser sensor are in data connection with a processor of the robot, and the processor is in data connection with a computer; and the application has the characteristics of effectively solving the problems of target inaccessibility and local minimum value of an artificial potential field method.

Description

Technical Field

[0001] This invention relates to the field of mobile robot technology, and in particular to a dynamic obstacle avoidance method for mobile robots based on a multi-feature repulsive potential field. Background Technology

[0002] With the development of technology and industrial transformation, mobile robots are widely used in various fields of production and life. Due to the uncertainty and complexity of the robot's working environment, how to quickly and accurately search for a collision-free path from the initial state to the target state has become a current technical challenge for robot obstacle avoidance.

[0003] Due to the different tasks of robots, the evaluation criteria for paths also vary. Generally, the best path is selected by evaluating criteria such as shortest path, shortest time, minimum energy consumption, and highest safety.

[0004] Currently, commonly used path planning algorithms are divided into two categories: artificial potential field methods and artificial potential field methods. The artificial potential field method assumes that the target point exerts an attractive force on the robot, while obstacles exert a repulsive force, causing the robot to move along the direction of downward force. The advantages of the artificial potential field method are its simple structure, which facilitates real-time control at the lower levels, significantly reduces computational load and time, and generates relatively smooth paths, thus helping to maintain robot stability.

[0005] Problems with the artificial potential field method:

[0006] 1. The problem of unreachable target: When there are large obstacles near the target point, the intelligent vehicle will be subjected to a great repulsive force when approaching the target, making it impossible to reach the target point.

[0007] Second, the local minimum problem is the bottleneck problem of the artificial potential field method in the zero potential energy field problem. The reason for the zero potential energy point is that there are multiple obstacle points in the working area of ​​the robot, and the location distribution is relatively special. At a certain point in the working area, the attraction and repulsion are exactly equal in magnitude and opposite in direction. The robot experiences zero force, and the robot will swing back and forth near that point and cannot reach the target point, which seriously affects the operating efficiency of the intelligent vehicle. Summary of the Invention

[0008] The purpose of this invention is to overcome the problems of target unreachability and local minima in the existing artificial potential field method, and to provide a dynamic obstacle avoidance method for mobile robots based on a multi-feature repulsive potential field.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A dynamic obstacle avoidance method for a mobile robot based on a multi-feature repulsive potential field, wherein the robot is equipped with a lidar, a positioning system, a three-dimensional laser sensor, and a four-line laser sensor; the lidar, positioning system, three-dimensional laser sensor, and four-line laser sensor are all connected to the robot's processor, and the processor is connected to a computer.

[0011] Includes the following steps:

[0012] Step 1: The lidar constructs a rectangular path planning map based on the environment, with rectangular walls around the edges of the path planning map;

[0013] Step 2: The positioning system obtains the robot's current position. The operator sets the target point position in the path planning map using a computer. The processor compares the current position with the target point position and determines whether the robot has reached the target point position.

[0014] Step 3: If the processor determines that the robot has reached the target point, path planning is completed; if the processor determines that the robot has not reached the target point, path planning is performed.

[0015] Preferably, if the processor determines that the robot has not reached the target point, path planning is performed according to the following steps:

[0016] Step 3-1: The 3D laser sensor and the four-line laser sensor detect and track dynamic obstacles, and the processor determines the distance between the robot and the target obstacle. The position, attitude, shape, velocity, and acceleration of each dynamic obstacle within;

[0017] Step 3-2, for each dynamic obstacle, the processor performs the following calculations:

[0018] 3-2-1, Let P be the gravitational potential field of the dynamic obstacle. erminal for:

[0019]

[0020] Where, k p (x) is the gravitational gain constant; robot ,y robot (x) represents the robot's current position. terminal ,y terminal (x) represents the robot's target point location; robot -x terminal ) 2 +(y robot -y terminal ) 2 This represents the square of the distance between the robot's current position and the target point.

[0021] Let P be the repulsive potential field of the dynamic obstacle. repulsive for:

[0022]

[0023]

[0024] Where, d obstacle The distance between the robot and the dynamic obstacle. η is the velocity of the dynamic obstacle. rep The repulsion coefficient is set.

[0025] 3-2-2, When both the robot and the dynamic obstacle are rectangular in shape, and the robot's width and length are 2a respectively. robot and 2b robot The width and length of the dynamic obstacle are 2a respectively. obstacle and 2a obstacle At that time, the point (X) on the edge of the rectangular machine. robot Y robot The equation for is expressed as:

[0026] The robot's left side: X robot cosθ robot -Y robot sinθ robot =-a robot ;

[0027] The robot's right side: X robot cosθ robot -Y robot sinθ robot =a robot ;

[0028] Above the robot: X robot sinθ robot +Y robot cosθ robot =-b robot ;

[0029] Below the robot: X robot sinθ robot +Y robot cosθ robot =b robot ;

[0030] Points (X) on the edge of the rectangular dynamic obstacle obstacle Y obstacle The equation for is expressed as:

[0031] Left side of the dynamic obstacle: X obstacle cosθobstacle -Y obstacle sinθ obstacle =-a obstacle ;

[0032] Left side of the dynamic obstacle: X obstacle cosθ obstacle -Y obstacle sinθ obstacle =a obstacle ;

[0033] Left side of the dynamic obstacle: X obstacle sinθ obstacle +Y obstacle cosθ obstacle =-b obstacle ;

[0034] Left side of the dynamic obstacle: X obstacle sinθ obstacle +Y obstacle cosθ obstacle =b obstacle ;

[0035] Where, θ robot θ represents the angle between the robot's direction of movement and the y-axis of the Cartesian coordinate system of the path planning map. obstacle This represents the angle between the moving direction of the dynamic obstacle and the y-axis of the Cartesian coordinate system of the path planning map. The moving direction of the dynamic obstacle is obtained from the attitude of the dynamic obstacle.

[0036] 3-2-3, When both the robot and the dynamic obstacle are elliptical in shape, and the lengths of the robot's major and minor axes are a... robot and b robot The lengths of the major and minor axes of the dynamic obstacle are a, respectively. obstacle and b obstacle When, then point (X) robot Y robot The equation for is expressed as:

[0037]

[0038] Among them, (x robot y robot () represents the center coordinates of the elliptical robot;

[0039] Then point (X) obstacle Y obstacle The equation for is expressed as:

[0040]

[0041] Among them, (x obstacle yobstacle () represents the center coordinates of the elliptical dynamic obstacle;

[0042] 3-2-4, The processor calculates the distance d between the dynamic obstacle and the robot. obstacle :

[0043]

[0044] Calculate the repulsive potential field of a rectangular wall and P wall :

[0045]

[0046]

[0047] in, d is the set safe distance between the robot and the rectangular wall. wall η is the distance between the robot and the rectangular wall. rep The repulsion coefficient is set.

[0048] 3-2-4-1, when the robot's length and width are 2a respectively robot and 2b robot When the rectangle is d wall for:

[0049]

[0050] 3-2-4-2, when the robot's major axis and minor axis lengths are respectively a robot and b robot When d is elliptical wall for:

[0051]

[0052] Where a is the width of the path planning map, b is the height of the path planning map; d1 is the distance between the robot and the right wall of the rectangular wall, d2 is the distance between the robot and the left wall of the rectangular wall, d3 is the distance between the robot and the top wall of the rectangular wall, and d4 is the distance between the robot and the bottom wall of the rectangular wall.

[0053] 3-2-5, Let P be the repulsive potential field of the dynamic obstacle. tend for:

[0054]

[0055] in, η represents the acceleration of a dynamic obstacle. trend The set trend coefficient;

[0056] The total potential field P acting on the robot is:

[0057] P = P terminal +P repulsive +P wall +P trend ;

[0058] The net force on the robot

[0059]

[0060] in, For Hamiltonian operators;

[0061] Then calculate the trend leader point p based on the sampling. steer :

[0062] p steer =p+L(F,t)

[0063] Where p represents the robot's current position, and L(F,t) represents the robot's position under the resultant force. Below, the distance moved after time t;

[0064] 3-3. Based on the force conditions at the trend guidance point, the processor determines whether the robot is about to fall into the problem of unreachable target and local minima.

[0065] If the processor encounters an unreachable target problem, it modifies the repulsion coefficient η. rep The processor can change the magnitude and direction of the net force acting on the robot at its current position to solve the goal-unreachable problem; if it gets stuck in a local minimum problem, the processor can add a random escape force. By changing the magnitude and direction of the net force acting on the robot at its current position, the robot can be freed from its equilibrium, thus solving the local minima problem.

[0066] Steps 3-4: The processor controls the robot to move based on the magnitude and direction of the various forces acting on the robot at its current position, and then returns to step 2.

[0067] Preferably, step 3-3 includes the following specific steps:

[0068] Step 3-3-1, Solve the problem of unattainable goal:

[0069] When the dynamic obstacle is close to the target point and much smaller When the trend-guiding point is determined to be in a goal-unreachable problem, meaning the robot is about to encounter such a problem, the processor reduces the repulsion coefficient η. rep This changes the magnitude and direction of the resultant force at the robot's current position;

[0070] Step 3-3-2, Solving the local minimum problem:

[0071] when When the trend-leading point is determined to be trapped in a local minimum problem, i.e., the robot is about to fall into a local minimum problem, the processor adds a random escape force to the robot. The resultant force and direction that change the current position.

[0072] Based on the traditional artificial potential field method, this invention sets up a trend repulsive potential field and a trend guidance point based on local sampling, according to the characteristics of obstacles such as speed and shape and the robot's own motion state. A trend coefficient is introduced into the trend repulsive potential field to improve the repulsive potential field. The trend guidance point is used to predict in advance whether the target is unreachable or local minima will occur, and to make path corrections, thereby fundamentally solving the target unreachability and local minima problems of the artificial potential field method.

[0073] Therefore, the present invention has the following beneficial effects: setting a trend repulsion potential field based on the characteristics of dynamic obstacles better ensures the safety of robot path planning; and effectively solves the problems of target unreachability and local minima in the traditional artificial potential field method. Attached Figure Description

[0074] Figure 1 This invention relates to a path planning map. Detailed Implementation

[0075] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0076] like Figure 1 The embodiment shown is a dynamic obstacle avoidance method for a mobile robot based on a multi-feature repulsive potential field. The robot is equipped with a lidar, a positioning system, a three-dimensional laser sensor, and a four-line laser sensor. The lidar, positioning system, three-dimensional laser sensor, and four-line laser sensor are all connected to the robot's processor, which is connected to a computer.

[0077] Includes the following steps:

[0078] Step 1: The lidar constructs a rectangular path planning map based on the environment, and a rectangular wall 1 is set on the edge of the path planning map;

[0079] Step 2: The positioning system obtains the current position of robot 2. The operator sets the target point position 4 in the path planning map through the computer. The processor compares the current position and the target point position and makes a judgment on whether the robot has reached the target point position.

[0080] Step 3: If the processor determines that the robot has reached the target point, path planning is complete; if the processor determines that the robot has not reached the target point, path planning proceeds as follows:

[0081] Step 3-1: The 3D laser sensor and the four-line laser sensor detect and track the dynamic obstacle 3, and the processor obtains the distance between the robot and the obstacle. The position, attitude, shape, velocity, and acceleration of each dynamic obstacle within;

[0082] For specific detection and tracking methods, please refer to the "Dynamic Obstacle Detection and Representation Method Based on Information Fusion of Three-Dimensional Laser Sensors and Four-Line Laser Sensors" published in Volume 36, Issue 6 of the journal Robotics.

[0083] Step 3-2, for each dynamic obstacle, the processor performs the following calculations:

[0084] 3-2-1, Let P be the gravitational potential field of the dynamic obstacle. terminal for:

[0085]

[0086] Where, k p (x) is the gravitational gain constant; robot ,y robot (x) represents the robot's current position. terminal ,y terminal () represents the robot's target point location;

[0087] (x robot -x terminal ) 2 +(y robot -y terminal ) 2 This represents the square of the distance between the robot's current position and the target point.

[0088] Let P be the repulsive potential field of the dynamic obstacle. repulsive for:

[0089]

[0090]

[0091] Where, d obstacle The distance between the robot and the dynamic obstacle. η is the velocity of the dynamic obstacle. rep The repulsion coefficient is set.

[0092] 3-2-2, When both the robot and the dynamic obstacle are rectangular in shape, and the robot's width and length are 2a respectively.robot and 2b robot The width and length of the dynamic obstacle are 2a respectively. obstacle and 2a obstacle At that time, the point (X) on the edge of the rectangular machine. robot Y robot The equation for is expressed as:

[0093] The robot's left side: X robot cosθ robot -Y robot sinθ robot =-a robot ;

[0094] The robot's right side: X robot cosθ robot -Y robot sinθ robot =a robot ;

[0095] Above the robot: X robot sinθ robot +Y robot cosθ robot =-b robot ;

[0096] Below the robot: X robot sinθ robot +Y robot cosθ robot =b robot ;

[0097] Points (X) on the edge of the rectangular dynamic obstacle obstacle Y obstacle The equation for is expressed as:

[0098] Left side of the dynamic obstacle: X obstacle cosθ obstacle -Y obstacle sinθ obstacle =-a obstacle ;

[0099] Left side of the dynamic obstacle: X obstacle cosθ obstacle -Y obstacle sinθ obstacle =a obstacle ;

[0100] Left side of the dynamic obstacle: X obstacle sinθ obstacle +Y obstacle cosθ obstacle =-b obstacle ;

[0101] Left side of the dynamic obstacle: X obstacle sinθ obstacle +Y obstacle cosθ obstacle =b obstacle ;

[0102] Where, θ robot θ represents the angle between the robot's direction of movement and the y-axis of the Cartesian coordinate system of the path planning map. obstacle This represents the angle between the moving direction of the dynamic obstacle and the y-axis of the Cartesian coordinate system of the path planning map. The moving direction of the dynamic obstacle is obtained from the attitude of the dynamic obstacle.

[0103] 3-2-3, When both the robot and the dynamic obstacle are elliptical in shape, and the lengths of the robot's major and minor axes are a... robot and b robot The lengths of the major and minor axes of the dynamic obstacle are a, respectively. obstacle and b obstacle When, then point (X) robot Y robot The equation for is expressed as:

[0104]

[0105] Among them, (x robot y robot () represents the center coordinates of the elliptical robot;

[0106] Then point (X) obstacle Y obstacle The equation for is expressed as:

[0107]

[0108] Among them, (x obstacle y obstacle () represents the center coordinates of the elliptical dynamic obstacle;

[0109] 3-2-4, The processor calculates the distance d between the dynamic obstacle and the robot. obstacle :

[0110]

[0111] Calculate the repulsive potential field of a rectangular wall and P wall :

[0112]

[0113]

[0114] in, d is the set safe distance between the robot and the rectangular wall. wall η is the distance between the robot and the rectangular wall. rep The repulsion coefficient is set.

[0115] 3-2-4-1, when the robot's length and width are 2a respectively robot and 2b robot When the rectangle is d wall for:

[0116]

[0117] 3-2-4-2, when the robot's major axis and minor axis lengths are respectively a robot and b robot When d is elliptical wall for:

[0118]

[0119] Where a is the width of the path planning map, b is the height of the path planning map; d1 is the distance between the robot and the right wall of the rectangular wall, d2 is the distance between the robot and the left wall of the rectangular wall, d3 is the distance between the robot and the top wall of the rectangular wall, and d4 is the distance between the robot and the bottom wall of the rectangular wall.

[0120] 3-2-5, Let P be the repulsive potential field of the dynamic obstacle. tend for:

[0121]

[0122] in, η represents the acceleration of a dynamic obstacle. trend The set trend coefficient;

[0123] The total potential field P acting on the robot is:

[0124] P = P terminal +P repulsive +P wall +P trend ;

[0125] The net force on the robot

[0126]

[0127] in, For Hamiltonian operators;

[0128] Then calculate the trend leader point p based on the sampling. steer :

[0129] p steer =p+L(F,t)

[0130] Where p represents the robot's current position, and L(F,t) represents the robot's position under the resultant force. Below, the distance moved after time t;

[0131] 3-3. Based on the force conditions at the trend guidance point, the processor determines whether the robot is about to fall into the problem of unreachable target and local minima.

[0132] If the processor encounters an unreachable target problem, it modifies the repulsion coefficient η. rep The processor can change the magnitude and direction of the net force acting on the robot at its current position to solve the goal-unreachable problem; if it gets stuck in a local minimum problem, the processor can add a random escape force. By changing the magnitude and direction of the net force acting on the robot at its current position, the robot can be freed from its equilibrium, thus solving the local minima problem.

[0133] Step 3-3-1, Solve the problem of unattainable goal:

[0134] When the dynamic obstacle is close to the target point and much smaller When the trend-guiding point is determined to be in a goal-unreachable problem, meaning the robot is about to encounter such a problem, the processor reduces the repulsion coefficient η. rep This changes the magnitude and direction of the resultant force at the robot's current position;

[0135] Step 3-3-2, Solving the local minimum problem:

[0136] when When the trend-leading point is determined to be trapped in a local minimum problem, i.e., the robot is about to fall into a local minimum problem, the processor adds a random escape force to the robot. The resultant force and direction that change the current position.

[0137] Steps 3-4: The processor controls the robot to move based on the magnitude and direction of the various forces acting on the robot at its current position, and then returns to step 2.

[0138] 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, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic obstacle avoidance method for mobile robots based on a multi-feature repulsive potential field, characterized in that, The robot is equipped with a lidar, a positioning system, a three-dimensional laser sensor, and a four-line laser sensor; the lidar, positioning system, three-dimensional laser sensor, and four-line laser sensor are all connected to the robot's processor, which is connected to a computer. Includes the following steps: Step 1: The lidar constructs a rectangular path planning map based on the environment, with rectangular walls around the edges of the path planning map; Step 2: The positioning system obtains the robot's current position. The operator sets the target point position in the path planning map using a computer. The processor compares the current position with the target point position and determines whether the robot has reached the target point position. Step 3: If the processor determines that the robot has reached the target point, then the path planning is complete. If the processor determines that the robot has not reached the target point, then path planning is performed: Step 3-1: The 3D laser sensor and the four-line laser sensor detect and track dynamic obstacles, and the processor determines the distance between the robot and the target obstacle. The position, attitude, shape, velocity, and acceleration of each dynamic obstacle within; Step 3-2, for each dynamic obstacle, the processor performs the following calculations: 3-2-1, Let P be the gravitational potential field of the dynamic obstacle. terminal for: Where, k p (x) is the gravitational gain constant; robot ,y robot (x) represents the robot's current position. terminal ,y terminal (x) represents the robot's target point location; robot -x terminal ) 2 +(y robot -y terminal ) 2 This represents the square of the distance between the robot's current position and the target point. Let P be the repulsive potential field of the dynamic obstacle. repulsive for: Among them, b obstacle The distance between the robot and the dynamic obstacle. η is the velocity of the dynamic obstacle. rep The repulsion coefficient is set. 3-2-2, When both the robot and the dynamic obstacle are rectangular in shape, and the robot's width and length are 2a respectively. robot and 2b robot The width and length of the dynamic obstacle are 2a respectively. obstacle and 2a obstacle At that time, the point (X) on the edge of the rectangular machine. robot Y robot The equation for is expressed as: The robot's left side: X robot cosθ robot -Y robot sinθ robot =-a robot ; The robot's right side: X robot cosθ robot -Y robot sinθ robot =a robot ; Above the robot: X robot sinθ robot +Y robot cosθ robot =-b robot ; Below the robot: X robot sinθ robot +Y robot cosθ robot =b robot ; Points (X) on the edge of the rectangular dynamic obstacle obstacle Y obstacle The equation for is expressed as: Left side of the dynamic obstacle: X obstacle cosθ obstacle -Y obstacle sinθ obstacle =-a obstacle ; Left side of the dynamic obstacle: X obstacle cosθ obstacle -Y obstacle sinθ obstacle =a obstacle ; Left side of the dynamic obstacle: X obstacle sinθ obstacle +Y obstacle cosθ obstacle =-b obstacle ; Left side of the dynamic obstacle: X obstacle sinθ obstacle +Y obstacle cosθ obstacle =b obstacle ; Where, θ robot θ represents the angle between the robot's direction of movement and the y-axis of the Cartesian coordinate system of the path planning map. obstacle This represents the angle between the moving direction of the dynamic obstacle and the y-axis of the Cartesian coordinate system of the path planning map. The moving direction of the dynamic obstacle is obtained from the attitude of the dynamic obstacle. 3-2-3, When both the robot and the dynamic obstacle are elliptical in shape, and the lengths of the robot's major and minor axes are a... robot and b robot The lengths of the major and minor axes of the dynamic obstacle are a, respectively. obstacle and b obstacle When, then point (X) robot Y robot The equation for is expressed as: Among them, (x robot y robot () represents the center coordinates of the elliptical robot; Then point (X) obstacle Y obstacle The equation for is expressed as: Among them, (x obstacle y obstacle () represents the center coordinates of the elliptical dynamic obstacle; 3-2-4, The processor calculates the distance d between the dynamic obstacle and the robot. obstacle : Calculate the repulsive potential field of a rectangular wall and P wall : in, d is the set safe distance between the robot and the rectangular wall. wall η is the distance between the robot and the rectangular wall. rep The repulsion coefficient is set. 3-2-4-1, when the robot's length and width are 2a respectively robot and 2b robot When the rectangle is d wall for: 3-2-4-2, when the robot's major axis and minor axis lengths are respectively a robot and b robot When d is elliptical wall for: Where a is the width of the path planning map, b is the height of the path planning map; d1 is the distance between the robot and the right wall of the rectangular wall, d2 is the distance between the robot and the left wall of the rectangular wall, d3 is the distance between the robot and the top wall of the rectangular wall, and d4 is the distance between the robot and the bottom wall of the rectangular wall. 3-2-5, Let P be the repulsive potential field of the dynamic obstacle. tend for: in, η represents the acceleration of a dynamic obstacle. trend The set trend coefficient; The total potential field P acting on the robot is: P=P terminal +P repulsive +P wall +P trend ; The net force on the robot in, For Hamiltonian operators; Then calculate the trend leader point p based on the sampling. steer : p steer =p+L(F,t) Where p represents the robot's current position, and L(F,t) represents the robot's position under the resultant force. Below, the distance moved after time t; 3-3. Based on the force conditions at the trend guidance point, the processor determines whether the robot is about to fall into the problem of unreachable target and local minima. If the processor encounters an unreachable target problem, it modifies the repulsion coefficient η. rep The processor can change the magnitude and direction of the net force acting on the robot at its current position to solve the goal-unreachable problem; if it gets stuck in a local minimum problem, the processor can add a random escape force. By changing the magnitude and direction of the net force acting on the robot at its current position, the robot can be freed from its equilibrium, thus solving the local minima problem. Steps 3-4: The processor controls the robot to move based on the magnitude and direction of the various forces acting on the robot at its current position, and then returns to step 2.

2. The method for dynamic obstacle avoidance of a mobile robot based on a multi-feature repulsive potential field according to claim 1, characterized in that, Step 3-3 includes the following specific steps: Step 3-3-1, Solve the problem of unattainable goal: If the trend-leading point is found to be stuck in the problem of unreachable target, the processor reduces the repulsion coefficient η. rep This changes the magnitude and direction of the resultant force at the robot's current position; Step 3-3-2, Solving the local minimum problem: when When the trend-leading point is determined to be trapped in a local minimum problem, i.e., the robot is about to fall into a local minimum problem, the processor adds a random escape force to the robot. The resultant force and direction that change the current position.

Images