Real-time RRT-DWO algorithm under dynamic and shape time-varying obstacle path planning environment

Abstract

The invention discloses a real-time RRT-DWO algorithm under a dynamic and shape time-varying obstacle path planning environment. The traditional path planning technology generally only considers the speed change of the obstacle, and the problem of introducing the new dynamic dimension of the shape change of the obstacle is solved. In order to cope with the complexity caused by shape change, the invention provides an RRT-DWA (Rapid-Extended Random Tree) algorithm based on an RRT (Rapid-Extended Random Tree) and a DWA (Dynamic Window Application) method, and develops a dynamic real-time RRT-DWA algorithm on the basis of the RRT-DWA algorithm so as to improve the real-time performance. In environments with different obstacle numbers and deformation rates, the dynamic real-time RRT-DWA algorithm has excellent performance in indexes such as success rate and average time, and has more obvious advantages especially under the condition that the deformation rate is accelerated. Figure 1 is an RRT algorithm sampling and searching process, and the algorithm is used as a core module for dynamic real-time RRT-DWA to obtain a global path.

Description

Technical Field

[0001] The invention belongs to the field of robot path planning, and in particular relates to a real-time RRT-DWO algorithm for solving path planning in a dynamic and time-varying obstacle environment. Background Art

[0002] Research on path planning can generally be divided into two main perspectives: the agent perspective and the obstacle perspective. From the agent perspective, research mainly focuses on single-agent and multi-agent scenarios. From the obstacle perspective, most works focus on static obstacles and dynamic obstacles. For dynamic obstacles, research mostly focuses on the case where the obstacle speed varies. However, the dynamics caused by the real-time change of obstacle shape has not received sufficient attention and exploration in existing research.

[0003] In fact, real-time changes in obstacle shapes are reflected in many application scenarios. In industrial automation, robots may encounter sudden changes in the shape of obstacles due to the stacking and movement of goods. In construction, the stacking of materials and the movement of equipment may cause changes in the shape of obstacles on site. When driving on temporary construction roads, autonomous vehicles need to respond in real time to changes in the impact range of obstacles caused by changes in shape. In logistics and distribution, unmanned delivery vehicles may face changes in the shape of obstacles due to the gathering of pedestrians. In disaster rescue scenarios, search and rescue robots must deal with constantly sinking and moving collapsed building structures and the changing shape of rubble piles.

[0004] In previous studies, static obstacles are usually handled using global path planning methods, such as the A* algorithm and the Rapidly-Exploring Random Tree (RRT). These methods are able to generate global paths, but have certain limitations when dealing with dynamic obstacles. Therefore, researchers usually use local path planning methods, such as the Artificial Potential Field (APF) and the Dynamic Window Approach (DWA) to deal with dynamic obstacles. Local path planning can achieve real-time obstacle avoidance, but the lack of global path guidance may cause the agent to fall into a local optimum and thus fail to reach the final goal.

[0005] Currently, many researchers have tried to combine global path planning with local path planning, that is, based on the global path generated by global path planning, local path planning is used for real-time obstacle avoidance. This method performs well when dealing with dynamic obstacles with only speed changes, but in complex scenarios where both speed and shape change dynamically, the probability of the agent successfully reaching the target is greatly reduced. To address this problem, our invention designs a method of real-time global path generation combined with local path update to meet the above challenges. Summary of the invention

[0006] The problem solved by the present invention is that, on the basis of coping with the change in speed of obstacles, we further cope with the newly added dynamic dimension of the dynamic factor of the real-time change in the shape of obstacles, and we further optimize the methods of global path planning and local path planning. Specifically, in terms of global path generation, we re-plan the global path of the agent in real time at each RRT node; at the same time, we perform real-time obstacle avoidance processing based on the DWA algorithm at each time step. This method can not only generate a safe global path in real time, ensuring that the agent moves safely towards the target in a highly dynamic environment, but also further enhance the agent's ability to respond to real-time changes in the environment.

[0007] The present invention is achieved through the following technical solutions:

[0008] The real-time RRT-DWO algorithm in the dynamic and time-varying obstacle path planning environment has the following specific steps:

[0009] Step 1: RRT algorithm generates global path

[0010] The RRT algorithm is mainly used to generate global paths. The basic idea of ​​generating global paths in a path planning environment is: take the starting point as the root node, randomly sample on the map, and use the nearest node as the parent node to generate corresponding child nodes, so that the random tree continues to expand. When the random tree approaches the end point and connects with the end point, the path search is completed.

[0011] Step 2: The DWA algorithm performs real-time obstacle avoidance based on the global path.

[0012] As a local path planning method, DWA can effectively help the agent to avoid obstacles in real time when facing dynamic obstacles. This method optimizes the linear velocity and angular velocity of the agent at each time step to achieve obstacle avoidance under the condition of dynamic changes of obstacles in the environment. The DWA algorithm evaluates each pair of velocities in the dynamic window through an objective function. This evaluation aims to measure the quality of the trajectory, including its alignment towards the target, the safe distance from the obstacle, and the magnitude of the velocity, with the goal of avoiding collisions while reaching the target quickly. The objective function J is defined as follows:

[0013] J=l1*J heading +l2*J dis +l3*J velocity

[0014] Among them, coefficients l1, l2 and l3 are weight coefficients.

[0015] Step 3: RRT-DWA algorithm

[0016] Introducing a new dynamic dimension, the time-varying obstacle shape, on the basis of the time-varying obstacle speed increases the unpredictability of the environment faced by the agent throughout the process. At the same time, the change in obstacle shape also exacerbates the damage to the path caused by obstacles during path planning. This highly dynamic environment places higher demands on the real-time response capability of the agent. In addition, the dynamic change of obstacles also leads to a significant expansion of the state space. Therefore, it is particularly important to provide global guidance in the initial stage to reduce the solution space and improve the path planning efficiency of the agent. We first designed Algorithm 1 to cope with the challenges of time-varying obstacle speed and shape during path planning.

[0017] Step 4: Dynamic Real-time RRT-DWA Algorithm

[0018] This algorithm adopts a more conservative safety strategy and re-plans the global path at each RRT node to generate a phased global safe path. On this basis, the DWA method is used for local path planning to ensure that the agent can generate a safe path in real time during the global path planning process. Compared with Algorithm 1, the dynamic real-time RRT-DWA algorithm has significantly improved safety and reliability. The specific implementation process is detailed in Algorithm 2.

[0019] Compared with the prior art, the present invention has the following beneficial technical effects:

[0020] The dynamic real-time RRT-DWA algorithm designed for path planning environments with dynamic and time-varying obstacles can play a key role in various scenarios such as industrial automation, construction, autonomous driving, logistics distribution and rescue operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is the sampling and searching process of RRT algorithm. This algorithm is the core module of dynamic real-time RRT-DWA to obtain the global path.

[0022] Figure 2 A conceptual diagram showing path planning under dynamic and time-varying obstacles.

[0023] Figure 3 (a) shows the path planning success rate results of the RRT algorithm under different numbers of obstacles and different deformation rates; Figure 3(b) shows the path planning success rate results of the DWA algorithm under different numbers of obstacles and different deformation rates; Figure 3 (c) shows the path planning success rate results of the RRT-DWA algorithm under different numbers of obstacles and different deformation rates; Figure 3 (d) shows the path planning success rate results of the dynamic real-time RRT-DWA algorithm under different numbers of obstacles and different deformation rates.

[0024] Figure 4 (a) shows the average path planning time results of the RRT algorithm under different numbers of obstacles and different deformation rates; Figure 4 (b) shows the average path planning time results of the DWA algorithm under different numbers of obstacles and different deformation rates; Figure 4 (c) shows the average path planning time results of the RRT-DWA algorithm under different numbers of obstacles and different deformation rates; Figure 4 (d) shows the average path planning time results of the dynamic real-time RRT-DWA algorithm under different numbers of obstacles and different deformation rates.

[0025] Figure 5 (a) shows the average distance results of path planning of the RRT algorithm under different numbers of obstacles and different deformation rates; Figure 5 (b) shows the average distance results of path planning of the DWA algorithm under different numbers of obstacles and different deformation rates; Figure 5 (c) shows the average distance results of path planning of the RRT-DWA algorithm under different numbers of obstacles and different deformation rates; Figure 5 (d) shows the average distance results of path planning of the dynamic real-time RRT-DWA algorithm under different numbers of obstacles and different deformation rates.

[0026] Figure 6 (a) shows the path planning results of the RRT algorithm when the deformation rate is 0.02 and the number of obstacles is 5, 10, 15, 20, and 25 respectively; Figure 6 (b) shows the path planning results of the DWA algorithm when the deformation rate is 0.02 and the number of obstacles is 5, 10, 15, 20, and 25 respectively; Figure 6 (c) shows the path planning results of the RRT-DWA algorithm when the deformation rate is 0.02 and the number of obstacles is 5, 10, 15, 20, and 25 respectively; Figure 6(d) shows the path planning results of the dynamic real-time RRT-DWA algorithm when the deformation rate is 0.02 and the number of obstacles is 5, 10, 15, 20, and 25 respectively. DETAILED DESCRIPTION

[0027] The present invention is further described in detail below in conjunction with specific embodiments, which are intended to explain the present invention rather than to limit it.

[0028] The real-time RRT-DWO algorithm in the dynamic and time-varying obstacle path planning environment has the following specific steps:

[0029] Step 1: RRT algorithm generates global path

[0030] The RRT algorithm is mainly used to generate global paths. The basic idea of ​​generating global paths in a path planning environment is: take the starting point as the root node, randomly sample on the map, and use the nearest node as the parent node to generate corresponding child nodes, so that the random tree continues to expand. When the random tree approaches the end point and connects with the end point, the path search is completed.

[0031] Step 2: The DWA algorithm performs real-time obstacle avoidance based on the global path.

[0032] As a local path planning method, DWA can effectively help the agent avoid obstacles in real time when facing dynamic obstacles. This method optimizes the linear velocity v and angular velocity ω of the agent at each time step to achieve obstacle avoidance under the condition of dynamic changes of obstacles in the environment. The DWA algorithm evaluates each pair of velocities in the dynamic window through an objective function. This evaluation aims to measure the quality of the trajectory, including its alignment J towards the target. heading , safe distance from obstacles J dis And the magnitude of the speed J velocity , the goal is to reach the target quickly while avoiding collision. The objective function J is defined as follows:

[0033] J=l1*J heading +l2*J dis +l3*J velocity

[0034] Among them, coefficients l1, l2 and l3 are weight coefficients.

[0035] This evaluation function evaluates the current state of the agent in real time during the local obstacle avoidance process, helping the agent to avoid obstacles as quickly and safely as possible, move in the direction of the target, and reach the target.

[0036] Step 3: RRT-DWA algorithm

[0037] Introducing a new dynamic dimension, the time-varying obstacle shape, on the basis of the time-varying obstacle speed increases the unpredictability of the environment faced by the agent throughout the process. At the same time, the change in obstacle shape also exacerbates the damage of obstacles to the path during path planning. This highly dynamic environment places higher demands on the real-time response capability of the agent. In addition, the dynamic change of obstacles also leads to a significant expansion of the state space. Therefore, it is particularly important to provide global guidance in the initial stage to reduce the solution space and improve the efficiency of the agent's path planning.

[0038] We first designed Algorithm 1 to address the challenges of time-varying speed and shape of obstacles during path planning.

[0039]

[0040] Step 4: Dynamic Real-time RRT-DWA Algorithm

[0041] In order to improve the responsiveness of the agent in a dynamic environment and ensure that it generates a safer path in a complex environment, we designed a dynamic real-time RRT-DWA algorithm. This algorithm adopts a more conservative safety strategy and re-plans the global path at each RRT node to generate a phased global safe path. On this basis, the DWA method is used for local path planning to ensure that the agent can generate a safe path in real time during the global path planning process. Compared with Algorithm 1, the dynamic real-time RRT-DWA algorithm has significantly improved safety and reliability. The specific implementation process is detailed in Algorithm 2.

[0042]

[0043]

[0044] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. Real-time RRT-DWO algorithm for path planning in dynamic and time-varying obstacle environment, characterized by: The specific steps are: Step 1: RRT algorithm generates a global path; Step 2: The DWA algorithm performs real-time obstacle avoidance based on the global path; Step 3: RRT-DWA algorithm; Step 4: Dynamic real-time RRT-DWA algorithm.

2. The real-time RRT-DWO algorithm for path planning in a dynamic and time-varying obstacle environment according to claim 1, characterized in that: The executed process can be described as follows: first, a feasible RRT path planning module is established, then the feasibility of the DWA module is verified, and on the basis of ensuring the feasibility of the RRT and DWA modules, a dynamic real-time RRT-DWA algorithm is constructed; the algorithm first initializes the RRT tree and generates a global path based on the RRT module; on this basis, a new global path is generated at each RRT node, and the DWA module is applied at each time step for local path planning.

3. The real-time RRT-DWO algorithm for path planning in a dynamic and time-varying obstacle environment according to claim 1, characterized in that: Based on traditional path planning technology, a new dynamic dimension, the change of obstacle shape, is introduced to solve the challenges brought by this dimension.

Images