Hybrid Sensor System Design Of Autonomous Cleaning Robot Based On The BP Neural Network And Obstacle Avoidance Planning

The sensor system, which is the "eyes" of the robot, is the foundation of robots' intelligence. So, to design a hybrid sensor system with perfect performance and low cost is the aim of this thesis. And the intelligent obstacle-avoiding algorithm based on BP neural network is designed as the "brain". All these are realized on the experimental platform of an autonomous cleaning robot.The obstacle information which is apperceived by the sensor system acts as input of the BP neural network, the output of the network acts as commands which drive the motors and the robot implement the obstacle avoiding behaviors according to the commands. The efficiency of the obstacle avoidance algorithm is validated with the experiments which are carried out on the robot experimental platform, which includes the DSP system for the algorithm processing and the MCU control system.This thesis includes four chapters, and the contents are as follows. The first chapter introduces the research status of the autonomous cleaning robot at home and abroad. Some classical sensor systems of the mobile robots and the cleaning robots are also introduced in this chapter. At the end, the main research content of this thesis is presented.In chapter 2, the hybrid sensor system with high reliability and perfect anti-jamming capability is designed considering the characteristic of sonar sensors and infrared sensors. And much experimental data about the characteristic of the designed sensor system is obtained and analyzed.The third chapter introduces the realization of the intelligent obstacle avoidance algorithm based on BP neural network. The designed BP neural network is simulated with MATLAB neural network toolbox, as well as the efficiency of the algorithm discussed in theory. And the optimal configuration of parameters is obtained through analyzing the performance parameters of the network.In chapter 4, the architecture and implementation of the hardware system is addressed. The platform includes the minimum application system for DSP, the MCU control system, the HPI communication between DSP and MCU, and the time-sequence matching between DSP and LCD by using CPLD.