Map Construction And Mobile Robot Navigation Based On Multi-sensor Information Fusion

With the continuous improvement of robotic technology,mobile robots gradually show their important uses in various industries.The main task of mobile robots is to sense the surrounding environment with sensors carried by themselves in unknown and complex environments.The environment map is constructed by sampling the environment by sensors,and the technology called Simultaneous Localization and Mapping(SLAM).At the same time,the development of data fusion technology makes it possible for robots to improve environmental awareness by using the fusion data of multi-sensors.Therefore,it is important for mobile robots in indoor navigation to use the multi-sensor information fusion technology.Firstly,FastSLAM algorithm of mobile robot is studied in this thesis.Aiming at the problem that the location error of mobile robot is large and the map accuracy is low when single sensor is used to estimate the position and pose of robot,an improved FastSLAM algorithm based on multi-sensor data fusion is proposed.The data of Inertial Measurement Unit(IMU)is fused with odometer information by using extended Kalman filter.And the fused information is used to replace the original odometer information to improve the accuracy of odometer model.Then the new odometer model and the laser observation model are used as the joint proposal distribution of FastSLAM to improve the accuracy.Finally,the method of resampling by FastSLAM particle filter is improved to slow down the particle degradation while maintaining the diversity of particles.Thus,a more accurate map is obtained.Secondly,the path planning methods of mobile robot are studied in this thesis.The navigation task of mobile robot is completed by using the combination of global path planning and local path planning methods.And the global path planning method is focused in this paper.In order to solve the problems that the traditional wolf swarm algorithm,which has single exploration direction,fixed running step and elimination rule of ‘survival of the fittest',that easily lead to slow convergence speed and falling into local optimal solution easily.I propose a global path planning algorithm based on improved wolf swarm algorithm.Firstly,the exploratory rules of fireworks algorithm is introduced to improve the global and local exploratory abilities.Secondly,the moving step of the runningbehavior is improved,so as to increase the efficiency of the running behavior.Finally,the diversity of wolf swarm is increased by improving the updating rules of wolf swarm algorithm,which can enhance the global optimization ability of the algorithm.The experiment of path planning for mobile robot is carried out by using the improved wolf swarm algorithm and the dynamic window algorithm.Experimental results show that the method proposed in this paper can quickly plan a global optimal path and avoid obstacles to reach the target position in actual operation process.Finally,a mobile robot navigation system based on map construction and information fusion is completed by building a mobile robot platform.Simulation and experiments show that the improved FastSLAM based on multi-sensor information fusion and improved resampling rules can output more accurate prior maps.Improved FastSLAM is more robust than the traditional FastSLAM.A high quality global path can be obtained by combining the global path planning algorithm of the improved wolf swarm algorithm with the local path planning algorithm.The feasibility and validity of the scheme are verified by experiments.Therefore,this paper focuses on improving the FastSLAM algorithm and the global path planning method of mobile robots,while taking into account the application of robots,which has a certain reference value.