Research On Service Robot's Navigation Under Indoor Environment

Navigation is one of the most hot and difficult problems of the mobile robot, as well as service robot, navigation is the prerequisites to complete a variety of services for service robot such as grabbing objects and walking aid , and the key to it's intelligent. This paper is based on "Service Robots for The Elderly" which is the main direction program of Chinese Academy of Sciences and the indoor service robot navigation problems were investigated, the paper includes the following aspects:First of all, introduced the history of service robots, the role and meaning for the Country, the role of navigation to the service robot related technologies, robot control and simulation platform of Player/Stage, and present research situation of home and abroad of service robot were described in detail.Second, for the indoor service robot working environment, this paper uses a static map as the model of the environment, and as a priori knowledge. Since the robot does not know it's initial position and orientation, we using a Monte Carlo Localization(MCL) based global positioning, MCL using the particle filter which using statistical-based sampling method to describe the probability distribution of the entire space, less samples for the small weighted particles and more samples for the higher weighted particles by re-sampling, control the number and distribution of particles, the algorithm does not accumulate errors over time, result in better positioning performance.In the local obstacle avoidance problem, we use the improved algorithm of VFH called VFH+, VFH+ using the four-stage data reduction instead of VFH two-stage reduction to calculate the new direction of movement, theory and experiment have shown that VFH+ is a good local obstacle avoidance algorithm, but under some conditions, robot may trap into the local dead-end, the paper then describes the VFH* algorithm which combines the A* algorithm to VFH+ to solve the problem.For global path planning problem, we use Wavefront global path planning algorithm, the robot using of waves expansion from the target moving towards the target point without collision. Using the Wavefront algorithm, the robot will not fall into any local minimum, and always choose the shortest path to reach the goal.In the Player/Stage platform, we implementation and simulation of the above algorithms, analyze the validity and apply them to real robot system.Finally, we summarize this paper; put forward its shortcomings, and present possible solutions.