1: \begin{abstract}

2: Many robotic tasks, such as human-robot interactions or the handling of fragile objects, require tight control and limitation of appearing forces and moments alongside sensible motion control \rev{to achieve safe yet high-performance operation}. We propose a learning-supported model predictive force and motion control scheme that provides \rev{stochastic} safety guarantees while adapting to changing situations. Gaussian processes are used to learn the uncertain relations that map the robot's states to the forces and moments. The model predictive controller uses these Gaussian process models to achieve precise motion and force control \rev{under stochastic constraint satisfaction}. 

3: \rev{As the uncertainty only occurs in the static model parts - the output equations - a computationally efficient stochastic MPC formulation is used. Analysis of} recursive feasibility of the optimal control problem and convergence of the closed loop system \rev{for the static uncertainty case are given}. \rev{Chance constraint formulation and back-offs} are constructed based on the variance of the Gaussian process to guarantee safe operation. The approach is illustrated on a lightweight robot \rev{in simulations and experiments}. % tasks while having contact with the environment. Simulation and experimental results underline the usability and safety of the  learning-supported predictive force-motion controller for unknown force models and disturbances.

4: \end{abstract}

5: