Research On Navigation Technology Of Indoor Mobile Robot Based On Laser And Vision

In recent years,with the deepening of development strategies such as "Industry4.0" and "Made in China 2025",the robot industry has also achieved rapid development.Autonomous navigation technology,as an important part of the intelligentization process of robot,has also become a hot issue of intelligent technology research in today.This thesis describes the current research status of robot navigation technology at home and abroad.Based on the analysis of the related technology of robot navigation system and common sensors,the laser rangefinder and RGB-D camera are selected as the main sensing equipments of the navigation system.A scheme of integrated navigation system for indoor mobile robot is proposed by establishing the system model.In the research of visual SLAM,this thesis mainly solves the problem of low accuracy of the visual sensor during the inter-frame registration and closed-loop detection.For the visual odometer module,this thesis establishes the visual odometer error model of the traditional ICP algorithm by mathematical method and introduces color vector to improve the traditional ICP algorithm.For the closed-loop detection module,the uncertainty of the camera in the spatial pose is found through the visual dictionary model,and is used as a constraint.Then,this constraint is introduced into the closed-loop detection algorithm.Finally,the best closed loop is found through the improved similarity score function.The experimental results show that the improved algorithm has higher registration accuracy and closed-loop detection accuracy.In the research of location and mapping algorithm,this thesis uses a method based on the unscented Kalman filter algorithm to fuse sensor attitude observation information to improve the accuracy of robot pose estimation.This method first uses the Bayesian probability model to design a fusion screening rule for observation information.Then,the appropriate sample points are selected for unscented transformation,and the mean and covariance of the transformed sample points are calculated.Next,the unscented Kalman filter algorithm is used to update and replace the mean value and covariance of the system.Finally,the posterior state estimation of the sample point is obtained,and the mapping is completed.The experimental results show that the improved algorithm has higher positioning accuracy.In the research of path planning,this thesis proposes a logarithmic ant colony algorithm based on potential field guidance to solve the problem of "blind" search and local optimization in the initial stage of ant colony algorithm.Based on a priori map of potential field model of the environment,the influence factors of potential field is introduced into the transition probability function of the ant colony algorithm and heuristic function,and through the logarithmic function model of ant colony algorithm improved pheromone update strategy.The path search algorithm on no longer has the blindness,the faster convergence speed and security.Experimental results show that the path planning algorithm proposed in this thesis has better path planning and shorter time.Finally,the indoor mobile robot navigation system are completed according to the fusion location mapping algorithm and path planning algorithm proposed in this thesis.The laboratory robot experiment platform is built to carry out map construction and path planning experiments respectively in the actual environment.The experimental results show that the indoor mobile robot integrated navigation system studied in this thesis is stable and feasible.