Dynamic Obstacle Avoidance For Mobile Robot Based On Multi–information Fusion

Firstly,the dynamic obstacle avoidance system is analyzed and studied,and the overall system design is completed.According to the design scheme,each hardware module and related software platform are elaborated.Through the analysis of the main research contents in this thesis,the focus of the research is to improve the fusion of external detection sensor information and dynamic obstacle avoidance algorithm.In this thesis,aiming at the problem of obstacle avoidance failure caused by the diversity of obstacles and the defect of the single sensor when detecting obstacles in the dynamic environment,a fusion method combining region fusion rule and Bayesian estimation is proposed for the detection data of laser sensor and an ultrasonic sensor to obtain more accurate environment.Information.The experimental results show that the method can complement the shortcomings of the two sensors and make the detection results more accurate and effective.At the same time,the internal sensor IMU,Hall encoder and external laser sensor are fused with extended Kalman filter to complete the real-time detection of mobile robot motion state.To overcome the shortcomings of the Q-learning dynamic obstacle avoidance algorithm in the process of updating Q value,an improved Q-learning dynamic obstacle avoidance algorithm based on neural network is proposed.The algorithm constrains the Q-value training process through a neural network,which makes the algorithm converge more easily and improves the security of obstacle avoidance.The experimental results show that the improved algorithm converges faster and makes the robot more efficient in obstacle avoidance.Finally,this thesis designs and constructs a dynamic obstacle avoidance system for the mobile robot based on multi-information fusion.By configuring the SLAM software platform based on ROS in the upper layer and debugging the embedded control board based on STM32 in the lower layer,the overall construction of a mobile robot is completed.Then,the performance of the overall dynamic obstacle avoidance method proposed in this thesis is validated by setting up a practical experimental scenario,including accuracy,security and effectiveness.The experimental results show that the proposed dynamic obstacle avoidance method for mobile robots based on multi-information fusion is more accurate and effective for environment detection,and the effect of obstacle avoidance is more safe and efficient.