Research On Indoor Autonomous Navigation Of Service Robot Based On Multi-sensor Information Fusion

With the continuous improvement of artificial intelligence,face recognition and voice interaction,the service robot industry presents a new trend of vigorous development.At the same time,driven by the increasing living standards of human beings,service robots are gradually showing the trend of family and intelligence.However,autonomous navigation technology,as the basis of Intelligent Implementation of service robots,still has some shortcomings.For example,in indoor environment,using single sensor for autonomous navigation can not get rich environmental information,low navigation accuracy and poor robustness.Therefore,the research on indoor autonomous navigation of service robots based on multi-sensor information fusion has important theoretical significance and practical application value.First of all,The status quo,navigation mode and sensor characteristics of service robots are studied in this thesis,the laser rangefinder,RGB-D depth sensor and physical odometer are used as the sensing devices of the service robot,and the software platform of robot operating system(ROS)is used to complete the design of the indoor autonomous navigation scheme of the service robot based on multi-sensor information fusion.Secondly,there are two ways to implement VO(visual odometry)in visual SLAM: direct method and feature method.Aiming at the problems that the direct method can not achieve closed-loop detection,the drift problem easily occurred in long time navigation and the time-consuming extraction and matching of feature method,a RGB-D vision SLAM algorithm which combines feature method with direct method is proposed.Based on the ORB-SLAM2 algorithm,the algorithm tracks non-key frames by minimizing photometric errors without feature extraction and matching;for key frames,the ORB(oriented fast and rotated brief)features are extracted and matched.Key frames are used for local mapping threads and closed loop threads,and non-key frames are used for tracking and positioning,which makes it less time-consuming.In addition,this algorithm can also build a complete and accurate 3D dense environment map based on RGB-D images and camera pose information,and provide more abundant environmental maps for service robots to complete navigation and obstacle avoidance tasks.Thirdly,in path planning,a path planning algorithm for service robots based on self-adaptive improvement ant colony optimization is proposed.The algorithm can solve the problems of slow convergence and easy to fall into local optimum in traditional ant colony algorithm,and the path is shorter and the convergence is better.The algorithm adopts parameter adaptive random transfer strategy in the initial stage of search,and introduces a security factor to improve the global search ability of the algorithm.Secondly,an heuristic factor is introduced into the transfer probability formula to avoid the algorithm falling into local optimum.Finally,the pheromone update rule is adjusted,and the pheromone volatilization factor is adjusted adaptively,so as to maintain a good balance between expanding search space and accelerating convergence speed.Experimental results show that the improved algorithm has better performance,faster convergence speed and optimal path planning.Finally,based on the vision SLAM algorithm and path planning algorithm proposed in this thesis,the design of indoor autonomous navigation system for service robot based on multi-sensor information fusion is realized.Through self built indoor service robot experimental platform,the visual SLAM and path planning algorithm are tested and analyzed.The experimental results show that the proposed algorithm is effective and feasible.The indoor autonomous navigation system of multi-sensor information fusion service robot researched in this thesis has good feasibility and reliability.