The Research On 3D Environment Perception System Under Uncertain Environment For Mobile Robot

This paper adopts Unknown Environment Perception System to perceive three-dimensional information of the environment, and studies on the accurate extraction of the target area for obstacles. The three-dimensional camera is applied for environment perception of mobile robot, which can acquire three-dimensional information of unknown environment in real time and reliably with high precision.This paper researches topography trafficability on complex environment and focuses on solving extraction of the target area for obstacles under complex scene and topography trafficability. In order to reduce unnecessary information interference and the amount of data to be analyzed, the paper applies characteristics of three-dimensional camera, and proposes the integration of three-dimensional information and gray scale information for the extraction of obstacle protuberance. By extracting the characteristics of obstacles, the paper identifies topography trafficability of unknown environment. The "threshold method" is applied to eliminate regions of non-obstacles which not only reduces unnecessary information interference but also reduces the amount of data to be analyzed. Thus, the computing speed is improved. Experiments show that the algorithm has relatively perception effects even under the condition of high consistence of the gray scale values of surrounding environment and obstacles.The obstacle region is extracted with the "threshold method". Then, the Euclidean distance is applied for cluster analysis on the remaining obstacle region. Based on the characteristics of extracted obstacles, the topography trafficability will be classified. For a more comprehensive study of extraction effects of different types of obstacles by three-dimensional perception system under unknown environment, several arbitrary complex scenes are selected for the test. Through tests in the common obstacles such as rocks, steep slopes, mixed complex region of rocks and steep slopes, obstacles affecting the robot movement in scenes are extracted and the specific properties of obstacles are measured. These have verified the proposed algorithm.This paper constructs the hardware platform of three-dimensional perception system, and the system is realzied through visual c++software. According to three-dimensional information of unknown environment, reconstruction of three-dimensional environment is completed. According to a number of complex scenes, types of obstacles are judged. The size and spacing of obstacles are also computed. This paper also analyzes three-dimensional topography trafficability under corresponding scenes so as to enable the precise guide and control of the system to give the accurate moves. It will also enable the mobile robot navigation fully adapt to any scene of unknown environment.The result of this paper optimizes the algorithm of the envrioment perception system, achieves real-time perception and analysis of mobile robot, and improves daptiveness of obstacle detection of the envrioment perception system in unknown environment. This paper integrates topographic three-dimensional information with the research of the environmental topography trafficability, presenting important theoretical values and practicability.