Research On Motion Control Of Slag Removal Robot Based On Vision And Force Perception

In the casting process of non-ferrous metallurgy,the dross produced by the oxidation of molten metal will be suspended on the metal surface.Therefore,removing the oxide slag is a key process affecting the quality of finished products.In the slag removal process,the oxidation slag is a semi-solid with time-varying stiffness,which is gradually cooled and solidified with time,and the spatial position of the oxidation slag also changes with the movement of the die.Based on this,this paper takes the slag removal robot as the research object,analyzes the surface characteristics of molten metal,and designs the motion control system of slag removal robot based on dynamics combined with visual and force sensing information,so as to provide technical support for the intelligent operation of slag removal robot.The main contents are as follows:(1)Aiming at the problem of time-varying stiffness of molten metal,an impedance control scheme with adaptive time-varying stiffness is designed.The time-varying contact force of molten metal is described by the combination of Maxwell model and oxidation growth model.An adaptive impedance control method based on reinforcement learning is designed to realize the dynamic correction of impedance parameters.Through comparative experiments,the feasibility of impedance control scheme based on Q-Learning is confirmed.(2)According to the demand of the slag raking robot for the actual control robustness in the unknown disturbance environment,the time delay estimation strategy is used to simplify the dynamic control model,and an adaptive fuzzy sliding mode control strategy combined with genetic algorithm optimization is proposed to ensure the stability of the control system.The adaptive ability and anti-interference ability of the control strategy are verified by experiments,and the control strategy meets the needs of industrial practical control.(3)Aiming at the limitation of slag removal robot in target perception,a control scheme based on vision and force perception information fusion is proposed.The front-end of the controller uses the Kalman filter algorithm of vision to optimize the dynamic trajectory.By introducing the visual situational memory mechanism into the controller,a control decision-making strategy based on memory comparison is designed to improve the decision-making efficiency of similar slag picking scenes.The control scheme is simulated in Coppeliasim software to verify that the scheme meets the actual needs of industry and ensure that the slag removal robot has the ability to deal with problems intelligently.