Research On Environment Perception Of Robot Based On Multi-sensor

Mobile robot is the important branch of robotics. It is high intelligent machine system that can sense the environment and their own state by sensors. Multi-sensor information Techniques is very important to build exact global map. The Pioneer 3-DX mobile robot produced by ActivMedia Corporation is taken as research platform to design and implement multi-object recognition, path planning, robot localization and environment perception.And, global landmark map is implemented.Firstly, the history and the development of the research of mobile robot are introduced. And machine vision which is used in mobile robot vision is discussed in detail.Secondly, this paper presents the hardware architecture and the software architecture of the Mobile Robot Pioneer 3-DX, the laser rangefinder, the monocular vision sensor and the sonar sensor. On the basis of the principle of robot motion, the locomotion of robot is analyzed.Next, the approaches of the color image processing and the multi-object recognition are presented, which include image segmentation, computing moment invariants and artificial neural network training to complete the multi-object recognition task.Finally, complete coverage path planning based on the grid that can make the robot to maximize coverage all the area is adopted. Moreover, as a local path planning algorithm robot can safely avoid obstacles by the method of using fuzzy logic. Then, the robot localization factors of uncertainty are recommended. At last, based on data of visual sensor and laser rangefinder is used to build global landmark map. The experiment results show that all above methods can build more exact global landmark map.