Robust NMPC Schemes for the Control of Mobile Robots in the Presence of Dynamic Obstacles

In recent years, there has been a large increase in the number of applications for mobile robots ranging from vacuum cleaners to Mars rovers. Depending on the applications, various challenges still need to be addressed related to the control of mobile robots and are being actively researched. We address here the problem of collision avoidance in the presence of dynamic obstacles using the Nonlinear Model Predictive Control (NMPC) framework. We use a scenario-tree formulation for the prediction of different trajectories of the dynamic obstacle and propose two strategies to avoid collisions. We propose a simple strategy to obtain the controlled inputs by solving a nominal NMPC problem by adding different scenarios of predicted obstacle trajectories as constraints. As a second strategy, a less conservative approach under the multi-stage NMPC framework by modeling the recourse in the predictions of the controlled robot is proposed. Both strategies can be applied in real-time and avoid collisions with dynamic obstacles. The non-conservative strategy shows superior performance at the expense of an increased computational effort.