Scenario-Based Trajectory Optimization in Uncertain Dynamic Environments

We present an optimization-based method to plan the motion of autonomous robot under uncertainties associated with dynamic obstacles, such as humans. Our bounds marginal risk collisions at each point in time by incorporating chance constraints into planning problem. This problem is not suitable for online optimization outright arbitrary probability distributions. Hence, we sample from these using uncertainty model, generate “scenarios,” which translate probabilistic deterministic ones. In practice, scenario represents collision constraint a obstacle location sample. The number theoretically required scenarios can be very large. Nevertheless, exploiting geometry workspace, show how prune most before and demonstrate reduced still provide guarantees on safety plan. Since our approach based, are able handle apply Model Predictive Contouring Control framework its benefits simulations experiments moving platform navigating among pedestrians, running real-time.