A Visual Servoing Method For Mobile Robots With Desired Pose Settings

With the continuous progress of control technology,wheeled mobile robots are more and more used in intelligent transportation,home service,industrial production and other fields.Vision sensor have the characteristics of low cost and no contact measurement for acquiring image information.In recent years,vision sensor combines with intelligent motion platform to form a visual servoing system of mobile robot,which has become a research hotspot in various fields.However,due to the nonholonomic constraints of the mobile robot itself and the uncertainty of the camera's internal and external parameters,the robot visual servoing system is still facing huge challenges.The nonlinear control method and vision based environment perception of the mobile robot have been studied deeply.In the existing control strategies,the desired pose and target feature points are mostly fixed.However,in the practical application of complex tasks,in the case of human set task requirements or external interference,the feature points and expected pose may change.Therefore,it is necessary to study the control method that the desired pose of the robot is adjusted reasonably with the movement of the feature points.This thesis mainly studies in the following aspects:1)A robot visual servoing method for the desired pose changing with feature points.This method can adjust the desired pose of the robot reasonably by moving the feature points,or keep the desired pose fixed relative to the feature points when the feature points are disturbed.In the task scene,four coplanar feature points with known coordinates are arbitrarily set.Then,in the case of target feature change,according to the coordinate system relationship of visual servoing system,the transformation relationship between the desired pose and current pose of robot is obtained by using the single homography decomposition and coordinate system transformation.Due to the random movement of feature points,the desired pose of in later transformation period is usually inconsistent with the robot's motion plane,so the adjusted desired pose of the robot is designed according to the projection principle.The polar coordinate control law is used to drive the robot from the current pose to target pose.Finally,the effectiveness of the method is analyzed according to the simulation and experimental results.2)A dynamic visual servoing method of mobile robot with unknown scene depth.In this thesis,a control strategy is designed to solve the problem that the desired pose of robot changes reasonably with the movement of feature points with unknown threedimensional coordinates.Add a fixed monitor camera to the scene.The image before and after the movement of the feature points captured by the robot's onboard camera and monitor camera.Then,using the six-degree-of-freedom homography matrix decomposition,the transformation relationship between the current pose and desired pose after the feature point moved is obtained.For the unknown depth existing in the transformation matrix,the two unknown depths are proportionally processed.The wheeled mobile robot is driven to the target pose by an adaptive controller.Simulation results and experimental results show that the method has good control performance.3)A visual servoing method between two desired poses based on wireless communication.Based on the last method,this thesis uses the wireless communication module to complete the visual servoing task from one desired pose to another desired pose for some practical tasks.After the analysis of several wireless communication modules,two XBee-Pro communication modules are selected,and the program is designed to send and receive data.The final desired position of the mobile robot of last method is set as the starting point.Through simulation and experimental results,the control strategy is effective to drive the mobile robot to the original desired pose before the feature point moves.