Autonomous grasping in dynamic environments where the object the robot or both are moving is an increasingly common task in robotic applications This thesis describes an online trajectory planner a geometric controller which evaluates a nonlinear memoryless function to map the current object position and velocity into a desired robot pose If the robot tracks these set points it is guaranteed to match the object s velocity and acceleration on a specied grasp surface A planar simulation demonstrates that this paradigm performs favorably when compared with the traditional planning approach Since it does not depend on future state predictions no object model is required Without trajectory prediction the computational effort is drastically reduced allowing for higher controller speed and tracking feedback gains The geometric controller has been successfully implemented on the dof Sarcos Dextrous Arm In order to provide the critical local sensing just before robotenvironment contact this thesis reports on work in progress towards the development of a proximity sensing network located in a robots multingered gripper This network will form an integral part of a multistage sensing system with vision and tactile sensor pads Sensing information is passed on to geometric controller which provides a framework for general sensorbased control of robotic tasks

Категория: Научные статьи | Нет комментариев »

Комментарии