Research And Simulation On Dynamic Shaking Compensation Method For Vision System Of Mobile Robot

While the mobile robot is moving, the vibration of vision system induced by uneven road and some other factors blur the captured image. It will bring tremendous difficulties to the following operations, such as target tracking and vision guidance. To solve the problem, humanoid and gorilla robot and its intelligent motion control lab, school of mechatronics engineering, harbin institute of technology has designed anti-shake device for vision system. In this thesis, based on the anti-shake device, related research on the dynamic shaking compensation is carried out to get steady camera image and improve the performance of the robot's vision system.First of all, the anti-shake device is improved according to the dynamic shaking compensation requirements by means of adding roll joint, which improves the anti-shake ability. At the same time, three-wheel mobile mechanism is designed, and the wheeled mobile robot virtual prototype with anti-shake device for vision system has been made.Secondly, dynamic shaking compensation method is proposed in the premise of no considering motion plan for each joint. Based on the shaking signals measured by sensors, equivalent transformation of the shaking is carried out. After that, the kinematics model of anti-shake device was used. And then the amount of shaking compensation for every joint is derived, which is regarded as the adjusted value for every joint.Afterwards, the dynamic shaking compensation control method was confirmed based on the characteristics of vision system anti-shake device, further dynamic analysis of the anti-shake device is carried out, and the co-simulation control system based on ADAMS and MATLAB was established by using the dynamic model.Finally, the dynamic shaking compensation motion simulation under artificial shaking environments was made by using designed anti-shake device for vision system, the result shows that shaking dynamic compensation can cut down the shaking of the camera sharply from the amplitude. The droop rate of peak to peak values of position and pose changes for camera is over seventy six percent. Simultaneously the dynamic shaking compensation motion simulation was made by using designed wheeled mobile robot with uneven virtual ground, the result shows that the dynamic shaking compensation method can reduce the shaking quantity of camera effectively and smooth the position and pose curves. From the results of two experiments above, it can be concluded that the dynamic shaking compensation method proposed in this paper is effective and feasible.