Research On SLAM Navigation Technology Of Indoor Service Robot Based On Fusion Of Laser And Vision

In recent years,with the development of robotics,the functions of service robots have become more and more intelligent.The autonomous navigation technology of mobile robots is the basis for achieving its intelligence,and autonomous robot navigation technology has gradually become one of the research hotspots in the field of robotics.This thesis describes the development of indoor service robots and the current research status at home and abroad,and studies the related technologies of robot navigation systems.After studying the characteristics of different navigation methods,the laser and vision fusion navigation system is selected.Subsequently,the observation model of laser and vision fusion is established,a fusion positioning algorithm is proposed,and a global and local fusion path planning algorithm is designed.Finally,based on the ROS(Robot Operating System)operating system,the design and implementation of the autonomous navigation system of indoor robots is completed.Aiming at the limitations of a single sensor for simultaneous localization and mapping,in order to improve the robot's global positioning accuracy,a laser and vision fusion localization algorithm is designed.Based on the AMCL(Adaptive Monte Carlo Localization)algorithm,this thesis studies the theoretical basis of sensor data fusion,and then builds a laser observation model and a visual observation model separately.The two observation model observe the map points independently of each other,so the product of the observation probabilities of the two is used as the posterior probability estimation of the robot pose by the fusion observation model,and the laser and vision fusion observation model is constructed based on this.The algorithm uses 60% of the particle weight as the lower limit.The particles are filtered by the laser beam observation model,and the visual observation model is used to assist in increasing the difference of the particles.High-resolution particles are more in line with real-world observations and are more scientific in resampling.Experimental results show that compared with AMCL algorithm,this algorithm has higher positioning accuracy and faster convergence speed.Aiming at the path planning problem,in order to improve the local obstacle avoidance ability of robot,a fusion path planning algorithm based on D* and DWA(Dynamic Window Approach)algorithm is proposed.The fusion path planning algorithm first generates a global path through the D* algorithm on the prior map,and checks the validity of the path in the DWA window;then divides the effective path into a series of nodes,and the DWA algorithm is used to generate local paths between nodes to realize dynamic obstacle avoidance.Finally,the fusion path planning algorithm is used to perform performance testing and simulation experiments to verify the reliability of the algorithm.Finally,this thesis completes the design of an indoor robot autonomous navigation system,which integrates the fusion SLAM(Simultaneous Localization and Mapping)algorithm and path planning algorithm proposed in this thesis.The navigation system is mounted on a robot experimental platform based on the ROS system,and then a dynamic and static obstacle environment navigation experiment is carried out.The experiment proves the feasibility and robustness of this system.