Perception and Manipulation in Robotics

The Perception and Manipulation Group aims to improve the sensing, planning and learning capabilities of robots to increase their autonomy and ease of use in manipulation tasks, especially for deformable objects. Some of the topics covered are the geometric interpretation of perceptual information (RGB-D vision, force and torque), the 3D modeling of rigid and flexible objects, action selection and planning, reinforcement learning, and teaching by demonstration.