Title:Path Integral Methods with Stochastic Control Barrier Functions

Abstract:Safe control designs for robotic systems remain challenging because of the difficulties of explicitly solving optimal control with nonlinear dynamics perturbed by stochastic noise. However, recent technological advances in computing devices enable online optimization or sampling-based methods to solve control problems. For example, Control Barrier Functions (CBFs), a Lyapunov-like control algorithm, have been proposed to numerically solve convex optimizations that determine control input to stay in the safe set. Model Predictive Path Integral (MPPI) uses forward sampling of stochastic differential equations to solve optimal control problems online. Both control algorithms are widely used for nonlinear systems because they avoid calculating the derivatives of the nonlinear dynamic function. In this paper, we utilize Stochastic Control Barrier Functions (SCBFs) constraints to limit sample regions in the sample-based algorithm, ensuring safety in a probabilistic sense and improving sample efficiency with a stochastic differential equation. We provide a sampling complexity analysis for the required sample size of our algorithm and show that our algorithm needs fewer samples than the original MPPI algorithm does. Finally, we apply our algorithm to a path planning problem in a cluttered environment and compare the performance of the algorithms.