Description of work
Handling large objects (for instance boxes, chairs) using a human/humanoid body offers several possible surfaces (torso, arms) for making contacts with the object. This whole body grasping/manipulation increases the number of contact points used to restrain the object and can provide a powerful way to handle large objects.
Based on previous ideas developed for grasp planning of enveloping (power) grasps with multifinger hands, the student will create and implement new strategies for handling large objects with humanoid robots, starting with the robot kinematics and a known model of the object to be manipulated. Feasibility and robustness of the grasp are important issues to consider during the planning phase. The approach will be initially implemented and validated using a virtual environment (OpenHRP or OpenRave), and later implemented on the humanoid robot Toro.
- Good knowledge of robot kinematics and dynamics
- Good knowledge of Matlab/Simulink and C++
- Previous knowledge of OpenHRP/OpenRave is desirable but not required
Negotiable, minimum 3 months in the case of internships, minimum 6 months in the case of master thesis