Research On Autonomous Robot Exploration

Robots have been widely used in industry,agriculture,military,etc.to ease the burden or replace human beings in heavy labor.As technology evolves,more and more demands are given to improve the intelligence and autonomy of robots.Especially in some harsh areas such as nuclear accidents,forest fires etc.,in order to carrying out investigation,exploration or rescue work,an environment map needs to be built firstly.An autonomous exploration system independent of external data needs to be designed for robot exploring unfamiliar environment and building map fully automatically.This thesis focuses on the topic of autonomous robot exploration,and proposes an autonomous robot exploration method including design and construction of both hardware layer and software layer,which includes map building,exploration approach and control method.The system builds point cloud map and 2D grid map of environment using a real-time mapping algorithm based on vision sensor and laser sensor,fuses position and orientation information from both sensors,makes exploration decisions based on the environment information in point cloud map and grid map,and leads the robot to explore.For flying robot such as a quad-rotor,in which the choice of on board sensors is limited because of weight and power supply,an autonomous exploration system based on fusion of vision sensors and IMU sensors is also proposed to accomplish the exploration of unfamiliar environment.Firstly,this thesis discusses about system framework design of autonomous robot exploration.A complete autonomous robot exploration method based on SLAM and data fusion of multiple sensors is proposed aiming at the application scenario using no external sensor.The system achieves real-time mapping and localization using data collected from vision sensor and laser sensor and gets more accurate localization result by fusing data of vision sensor,laser sensor and odometer.The system also controls robot to move and to accomplish autonomous exploration according to map and exploration strategy.Secondly this thesis focuses on exploration algorithm.The Ambient Environment Based Exploration Approach is proposed.According to known and unknown environment information in the built map,the approach works out a strategy to explore unknown area in the map.This thesis shows the algorithm and analyzes its characteristics.Then the thesis introduces the design of both hardware and software platform for autonomous robot exploration.This part introduces composition,design and construction of motion control system,taking an intelligent vehicle as example.A construction method is introduced for sensor drivers,data processing and motion control system based on Linux and ROS.Finally the process of Ambient Environment Based Approach is simulated on computer.A great amount of simulation results are achieved which verifies the effectiveness of proposed autonomous exploration strategy.Through the experiments of intelligent vehicle and flying robot,the function of autonomous robot exploration method in unfamiliar environment is also verified.