| Home | People | Van der Loos | Student Projects - 2005ProVAR, Integrating Collision Avoidance and Contact Management When designing a robot system, one of the most important things to implement is a controller that can move the manipulator to a desired location. To make this controller work properly under different conditions, the inertial, centrifugal, Coriolis and gravity forces that affect the manipulator when in motion, also called the dynamics, must be taken into account. After implementing this controller, the robot must be able to interact with its environment, i.e., perform certain pre-programmed tasks like pushing a button, turning a switch or placing a cup on the desk. This interaction with objects and devices involves "contact management". Hence contact management can be considered as the opposite of "collision avoidance", which is used to avoid contact at all costs. Being able to detect obstacles and avoid collisions is of importance for a robot working in the presence of humans. A significant amount of research has been done and is still proceeding with the goal of increasing the safety of the interaction with the environment and with people, leading to much more varied applications of assistant robots than previously imaginable. Our work has been focused on local collision avoidance in real time, integrating formulas developed in the 1980s by Stanford Professor Oussama Khatib and his students. Sensors mounted on the manipulator have been used to collect data from the environment. Our major contribution has been to write the algorithms to implement robust, safe, sensor-based trajectory control for ProVAR. Driver's SEAT, Generation II |