Research On Sharing Collaboration System Of Robot Based On Vision Guidance

The development of Robotics is at a critical stage.It is foreseen that robots will occupy all areas of life and play an active role.However,judging from current development status,we need to clearly recognize that the development level of controlling technologies,sensors,and artificial intelligence limits the robot development.In the short term,it is still a long way to develop a robot that can be fully autonomous in a complex and time-varying environment.When a robot performs complex control tasks,joining human's control behavior,that is,incorporating highlevel decision-making of people in task control can circumvent some technical obstacles that still exists in the robot control.From the perspective of humans,the robot's work content meeting the requirements of people at the right time,people and robots working together to accomplish specific tasks,reduce people's work intensity and improve people's working conditions.In this paper,a visual guided human-machine cooperative control system is designed to achieve human-computer cooperative control.The study combines visual guidance with visual servoing that has more universal applicability under a variety of working conditions.The visual servo has good robustness in the local area.However,existing studies on the visual servo control,the robot in a time-varying and complex environment,there are still problems in global divergence and difficult to achieve obstacle avoidance.In order to solve this kind of problem,when constructing the human-computer collaboration control system,people's advanced decision-making is integrated into the control system,relying on humans to carry out vision-aided manipulators to achieve obstacle avoidance.The end of the mechanical arm is visually guided to the vicinity of the target object,and taking the good convergence characteristics advantage of the image visual servoing control in the vicinity of the target object,realizing accurate servo positioning of the target feature specified information,thereby completing the man-machine cooperative control task.The research includes two parts: manipulator arm visual guidance control and image visual servo control.The following work is accomplished by constructing the human-computer collaborative control system:1.Hand vision guide(1)Multiple visual sensors layout was proposed.Aiming at visual guidance,the spatial layout of multiple visual sensors was proposed for visual guidance.According to the characteristic of Kinect data,neural network was applied to achieve multi-sensor data fusion.(2)Two methods for calculating spatial relationships were proposed.Firstly,based on the platform identification method of spatial relationship determination,replaced the previous calibration methods to get the conversion relationship between the camera space and the manipulator working space.The working plane recognition and reconstruction were used to construct the mapping relationship of the visual sensorrobot working space.The second method was based on the spatial relationship determination method of point cloud template matching.Through the pre-sampling of the three-dimensional geometric model,a single view point cloud template library of the robot arm was built to match with the point cloud collected at the scene to determine the spatial mapping relationship.(3)Visual guidance control was established.Taking the positional change of the skeletal point of the hand as the position control information of the end of the manipulator,the Kalman filter is used to track and reduce the noise,the non-uniform B-spline curve performed trajectory smoothing processing.The hand point cloud information is used as the end posture control information of the robot arm.The hand point cloud is segmented and noise reduction processing to be performed.In addition,a visual guided simulation platform was constructed and visual guided simulation experiments with hand control information as input were performed.2.Image vision servo(1)Visual servo model was constructed.Hand-eye relationship and cameraspecific parameters were determined.The study achieved the target object's feature recognition and positioning.A solution to improve the efficiency of visual servo control was proposed,that is,an adaptive gain strategy to achieve efficient control of visual servoing.The control law was adjusted to achieve continuous speed control of the arm.(2)The task switching strategy from visual guidance to visual servo control was established.The study constructed the visual guidance-visual servo intelligent smooth transition strategy based on template matching.In addition,visual servo control simulation experiment and simulation results analysis were performed.The study established a visual guidance-visual servo human-robot cooperative control system.A control strategy from visual guidance transition to visual servoing was designed to achieve a smooth and intelligent transition.During the control period,the operator feel convenient and flexible in operation.The control system is stable and reliable robust.