Autonomous Robotic Exploration Based On Frontier Point Optimization And Multistep Path Planning

Autonomous robotic exploration of an unknown environment is a key technology for robot intelligence.The primary goal is to make robots to acquire the most complete and accurate map of an environment in a finite time without human intervention.At present,many existing map exploration strategies are based on frontier.In order to improve the efficiency of it,we propose a strategy based on frontier point optimization and multistep path planning in this thesis.The entire exploration strategy is divided into two major parts: the determination of the target frontier point and the path planning of the robot.For these two parts,this thesis mainly carried out the following works:On the problem of determining the target point,the optimal frontier point extraction algorithm is proposed.We first apply the Rapidly-exploring Random Tree(RRT)algorithm to frontier points generation.The evaluation function of frontier points is defined by considering information gain,navigation cost and the precision of the localization of the robots.On this basis,we borrow the idea of Glowworm Swarm Optimization(GSO)algorithm to optimize the generated frontier points and select the frontier point with the highest evaluation value as the target frontier point.In the path planning section,we propose a multistep exploration strategy.First,we improved the trajectory evaluation function of the dynamic window approach(DWA).Instead of planning the global path from the current position of the robot to the target frontier point directly,we set a local exploration path step size.When the robot's movement distance reaches the local exploration path step size,we reselect the current optimal frontier point for path planning to reduce the possibility that the robot may take some repetitive paths.Finally,the relevant experiments are carried out to verify the effectiveness of this strategy.