Research on Mobile Robot Target Recognition and Obstacle Avoidance Based on Vision

Abstract

This work investigates the method of object recognition and autonomous obstacle avoidance for mobile robot based on vision, and solves to the problem that mobile robot can move and identify objects in complex environment. Build the Mecanum wheel mobile platform and equip the Kinect sensor and laser rangefinder sensor, transmitted data to the host computer to information decision analysis, and the execution system generation mobile robot motion control commands, then upload the commands to the cloud data center and establish the SQL Server database table to store the information. Robot scan 100ms of every table, to perform the latest control information. 0bject detection uses Gauss model background difference method to detect objects. Target feature extraction adopts the SURF (Speed Up Robust Features) algorithm and uses RANSAC (Random Sample Consensus) algorithm to optimization, removing the mismatching points improves the computation speed and the detection accuracy. Autonomous obstacle avoidance module through the laser range finder to collect distance information, set the safety distance with the obstacles, can achieve the effect of self-obstacle avoidance. The experiment proves that the robot can complete the task of target recognition and autonomous obstacle avoidance, and verify the effectiveness of the system.