2P2-G05 Path Searching of Mobile Robot for Obstacle Avoidance Using Reinforcement Learning : Integration of A* algorithm and Reinforcement Learning

2008 ◽  
Vol 2008 (0) ◽  
pp. _2P2-G05_1-_2P2-G05_3
Author(s):  
Kenji HIRAOKA ◽  
Seiji AOYAGI
Keyword(s):  
Reinforcement Learning ◽  
Mobile Robot ◽  
Obstacle Avoidance ◽  
A Algorithm

scholarly journals Research on path planning of three-neighbor search A* algorithm combined with artificial potential field

2021 ◽  
Vol 18 (3) ◽  
pp. 172988142110264
Author(s):  
Jiqing Chen ◽  
Chenzhi Tan ◽  
Rongxian Mo ◽  
Hongdu Zhang ◽  
Ganwei Cai ◽  
...  
Keyword(s):  
Path Planning ◽  
Mobile Robot ◽  
Obstacle Avoidance ◽  
Potential Field ◽  
Path Length ◽  
Search Time ◽  
Experimental Results ◽  
Artificial Potential Field ◽  
A Algorithm ◽  
Neighbor Search

Among the shortcomings of the A* algorithm, for example, there are many search nodes in path planning, and the calculation time is long. This article proposes a three-neighbor search A* algorithm combined with artificial potential fields to optimize the path planning problem of mobile robots. The algorithm integrates and improves the partial artificial potential field and the A* algorithm to address irregular obstacles in the forward direction. The artificial potential field guides the mobile robot to move forward quickly. The A* algorithm of the three-neighbor search method performs accurate obstacle avoidance. The current pose vector of the mobile robot is constructed during obstacle avoidance, the search range is narrowed to less than three neighbors, and repeated searches are avoided. In the matrix laboratory environment, grid maps with different obstacle ratios are compared with the A* algorithm. The experimental results show that the proposed improved algorithm avoids concave obstacle traps and shortens the path length, thus reducing the search time and the number of search nodes. The average path length is shortened by 5.58%, the path search time is shortened by 77.05%, and the number of path nodes is reduced by 88.85%. The experimental results fully show that the improved A* algorithm is effective and feasible and can provide optimal results.

A Task of Miniature Mobile Robot Learning for Obstacle Avoidance through Neural Networks

2012 ◽  
Vol 151 ◽  
pp. 498-502
Author(s):  
Jin Xue Zhang ◽  
Hai Zhu Pan
Keyword(s):  
Neural Networks ◽  
Reinforcement Learning ◽  
Mobile Robot ◽  
Obstacle Avoidance ◽  
Robot Learning ◽  
Q Learning ◽  
Robot Systems ◽  
The Neural Networks ◽  
Behavior Based

This paper is concerned with Q-learning , a very popular algorithm for reinforcement learning ,for obstacle avoidance through neural networks. The principle tells that the focus always must be on both ecological nice tasks and behaviours when designing on robot. Many robot systems have used behavior-based systems since the 1980’s.In this paper, the Khepera robot is trained through the proposed algorithm of Q-learning using the neural networks for the task of obstacle avoidance. In experiments with real and simulated robots, the neural networks approach can be used to make it possible for Q-learning to handle changes in the environment.

scholarly journals Multimodal Deep Reinforcement Learning with Auxiliary Task for Obstacle Avoidance of Indoor Mobile Robot

Sensors ◽  
2021 ◽  
Vol 21 (4) ◽  
pp. 1363
Author(s):  
Hailuo Song ◽  
Ao Li ◽  
Tong Wang ◽  
Minghui Wang
Keyword(s):  
Reinforcement Learning ◽  
Mobile Robot ◽  
Obstacle Avoidance ◽  
Representation Learning ◽  
Velocity Estimation ◽  
Depth Images ◽  
Indoor Mobile Robot ◽  
Multimodal Information ◽  
Testing Environments ◽  
Accumulated Reward

It is an essential capability of indoor mobile robots to avoid various kinds of obstacles. Recently, multimodal deep reinforcement learning (DRL) methods have demonstrated great capability for learning control policies in robotics by using different sensors. However, due to the complexity of indoor environment and the heterogeneity of different sensor modalities, it remains an open challenge to obtain reliable and robust multimodal information for obstacle avoidance. In this work, we propose a novel multimodal DRL method with auxiliary task (MDRLAT) for obstacle avoidance of indoor mobile robot. In MDRLAT, a powerful bilinear fusion module is proposed to fully capture the complementary information from two-dimensional (2D) laser range findings and depth images, and the generated multimodal representation is subsequently fed into dueling double deep Q-network to output control commands for mobile robot. In addition, an auxiliary task of velocity estimation is introduced to further improve representation learning in DRL. Experimental results show that MDRLAT achieves remarkable performance in terms of average accumulated reward, convergence speed, and success rate. Moreover, experiments in both virtual and real-world testing environments further demonstrate the outstanding generalization capability of our method.

scholarly journals A New Hybrid Method in Global Dynamic Path Planning of Mobile Robot

2018 ◽  
Vol 13 (6) ◽  
pp. 1032-1046 ◽  
Author(s):  
Xiaoru Song ◽  
Song Gao ◽  
C.B. Chen ◽  
Kai Cao ◽  
Jiaoru Huang
Keyword(s):  
Path Planning ◽  
Mobile Robot ◽  
Real Time ◽  
Obstacle Avoidance ◽  
A Algorithm ◽  
Fusion Algorithm ◽  
Global Dynamic ◽  
Dynamic Path Planning ◽  
Dynamic Path ◽  
Dynamic Obstacle

Path planning and real-time obstacle avoidance is the key technologies of mobile robot intelligence. But the efficiency of the global path planning is not very high. It is not easy to avoid obstacles in real time. Aiming at these shortcomings it is proposed that a global dynamic path planning method based on improved A* algorithm and dynamic window method. At first the improved A* algorithm is put forward based on the traditional A* algorithm in the paper. Its optimized heuristic search function is designed. They can be eliminated that the redundant path points and unnecessary turning points. Simulation experiment 1 results show that the planned path length is reduced greatly. And the path transition points are less, too. And then it is focused on the global dynamic path planning of fusion improved A* Algorithm and Dynamic Window Method. The evaluation function is constructed taking into account the global optimal path. The real time dynamic path is planning. On the basis of ensuring the optimal global optimization of the planning path, it is improved that the smoothness of the planning path and the local real-time obstacle avoidance ability. The simulation experiments results show that the fusion algorithm is not only the shorter length, but also the smoother path compared the traditional path planning algorithms with the fusion algorithm in the paper. It is more fit to the dynamics of the robot control. And when a dynamic obstacle is added, the new path can be gained. The barrier can be bypass and the robot is to reach the target point. It can be guaranteed the global optimality of the path. Finally the Turtlebot mobile robot was used to experiment. The experimental results show that the global optimality of the proposed path can be guaranteed by the fusion algorithm. And the planned global path is smoother. When the random dynamic obstacle occurs in the experiment, the robot can be real-time dynamic obstacle avoidance. It can re-plan the path. It can bypass the random obstacle to reach the original target point. The outputting control parameters are more conducive to the robot’s automatic control. The fusion method is used for global dynamic path planning of mobile robots in this paper. In summary the experimental results show that the method is good efficiency and real-time performance. It has great reference value for the dynamic path planning application of mobile robot.

scholarly journals Mobile Robot Obstacle Avoidance Based on Deep Reinforcement Learning

2019 ◽  
Author(s):  
Shumin Feng ◽  
Hailin Ren ◽  
Xinran Wang ◽  
Pinhas Ben-Tzvi
Keyword(s):  
Mathematical Model ◽  
Reinforcement Learning ◽  
Mobile Robot ◽  
Learning Strategies ◽  
Obstacle Avoidance ◽  
Autonomous Navigation ◽  
Sensor Data ◽  
Experience Replay ◽  
Testing Environments ◽  
Avoidance Problem

Abstract Obstacle avoidance is one of the core problems in the field of mobile robot autonomous navigation. This paper aims to solve the obstacle avoidance problem using Deep Reinforcement Learning. In previous work, various mathematical models have been developed to plan collision-free paths for such robots. In contrast, our method enables the robot to learn by itself from its experiences, and then fit a mathematical model by updating the parameters of a neural network. The derived mathematical model is capable of choosing an action directly according to the input sensor data for the mobile robot. In this paper, we develop an obstacle avoidance framework based on deep reinforcement learning. A 3D simulator is designed as well to provide the training and testing environments. In addition, we developed and compared obstacle avoidance methods based on different Deep Reinforcement Learning strategies, such as Deep Q-Network (DQN), Double Deep Q-Network (DDQN) and DDQN with Prioritized Experience Replay (DDQN-PER) using our simulator.

Sign in / Sign up

Export Citation Format

Share Document