Research On Mobile Robot Path Planning Method Based On Deep Reinforcement Learnin

As an important research direction of mobile robot,path planning means that the robot can effectively avoid obstacles from the starting point to the target point and independently plan a reasonable path.It is the key to realize the navigation task.Traditional path planning algorithms have made remarkable achievements in path planning under known map environment.But there is still a lot of research space under unknown map environment.The deep reinforcement learning algorithm can solve this problem by using the exploration and interactive learning between robot and environment.Deep reinforcement learning algorithm perceives the environment through deep learning.It uses reinforcement learning to make decisions and judgments on the perceived information.Therefore,aiming at the path planning problem of mobile robot in unknown environment,an improved deep reinforcement learning algorithm based on depth image information is proposed for mobile robot path planning.By using Kinect V2 visual sensor,the depth image information directly collected is taken as the input of the Deep Q Network.The speed of robot is used as the output of the action information.By introducing TOF ranging principle to measure the distance between robot and obstacles,the state action space is optimized.The instability of discrete motion of robot is solved.The linear velocity and angular velocity of the robot are designed.The reward and punishment function is improved.The problem of sparse reward is alleviated.The Q-value function is introduced to improve the action selection strategy and balance the relationship between exploration and utilization.The robot outputs better actions,and its autonomous exploration and driving speed is accelerated.The simulation environment of different paths is designed,and the effect of the improved algorithm is verified.The physical test of mobile robot based on Turtlebot2 is carried out.The efficiency of path planning is improved.The effectiveness of the improved DQN algorithm in path planning of mobile robot is verified.