A challenging mechatronics problem is the control of a robot arm equipped with a dextrous hand for skillful manipulation tasks. The position and velocity of the robot joints and force in the arm and fingers need to be controlled using information from the robot's position, force and vision sensors. The control is effected by torque control on the servo motors in the joints of the robot and the fingers in the robot hand. The problem here is one of dealing with essential nonlinearities, large data throughput from the sensors and low signal to noise ratio from the force sensors. One task is to achieve optimal balancing of grasping forces in a dynamic situation to avoid slippage and excessive force and allow manipulation of objects. Another key problem is to develop the real-time computational methods to control the dynamic motion models for the entire robotic system. A three fingered hand which has 3 degrees of freedom (DOF) in each finger that is mounted on a 7 DOF robot arm represents an 16 DOF system. Our ultimate goal is to mount the robot arm-hand on a mobile platform (yielding a 19 DOF system) which will allow research of the performance of skillful tasks such as opening and passing through doors. Such tasks require the coordination and the real-time control of all the robots actuators; fingers, arm and wheels. We intend to use this robot with people as a set of helping hands. This will require human-robot communication interfaces and the development of safety mechanisms that protect the person from injury. In this project we are using a Barrett whole arm manipulator (WAM) robot with 7DOF together with Barrett's three (3) fingered hand as a platform for research in dextrous manipulation. We have also acquired a Nomadic Technologies Robot which can cappy the Barrett arm and thus allows us to conduct research into the problems of mobile manipulators.
| RSL | PROJECTS | PAPERS | PEOPLE | DEMOS | EVENTS | LABS |