Robotics

Robots should move and interact with humans as efficiently as possible. If this includes high velocities or heavy payloads, the nonlinear rigid body dynamics as well as the input constraints need to be taken into account, which leads to a computationally demanding optimization problem. On the other hand, when the robot is in contact with its environment, the rigid body dynamics are often less relevant, but the forces and torques that arise must be considered by the control system. This is especially important for safe human-robot interaction. Here, the challenges include the contact modelling (e.g. stiffness and friction), the safe handling of contact loss, or the selection of controller parameters for different applications. Current research considers model predictive interaction control (MPIC), which refers to MPC with explicit prediction of contact forces and torques, as well as the development of specialized algorithms for solving optimization problems with rigid body dynamics.

Besides controlling motions, also the planning of paths (geometric description) and trajectories (time information) is a relevant problem for many types of robots. In particular for mobile and collaborative robots, motions should be planned in such a way that they do not cause self-collisions or collisions with obstacles in the environment. Global planners iteratively build a search structure that explores the space of possible motions, whereas local planners only search in the neighborhood of an initial solution. In order to efficiently find high-quality solutions, it is necessary to combine the advantages of both global and local planning methods. Further difficulties arise in dynamic environments, where the future motion of obstacles needs to be predicted or for car-like robots, where the non-holonomic kinematics need to be considered.

Videos

Related projects

Related publications