CoVO-MPC: Theoretical Analysis of Sampling-based MPC and Optimal Covariance Design
This work addresses a theoretical gap in MPC for robotics and reinforcement learning, offering significant performance gains in agile control tasks.
The paper tackles the lack of theoretical understanding in sampling-based Model Predictive Control (MPPC), specifically analyzing the convergence of Model Predictive Path Integral Control (MPPI) and showing it achieves at least linear convergence rates for quadratic optimizations. It introduces CoVo-MPC, a novel algorithm that optimally schedules sampling covariance to improve convergence, empirically outperforming standard MPPI by 43-54% in simulations and real-world quadrotor tasks.
Sampling-based Model Predictive Control (MPC) has been a practical and effective approach in many domains, notably model-based reinforcement learning, thanks to its flexibility and parallelizability. Despite its appealing empirical performance, the theoretical understanding, particularly in terms of convergence analysis and hyperparameter tuning, remains absent. In this paper, we characterize the convergence property of a widely used sampling-based MPC method, Model Predictive Path Integral Control (MPPI). We show that MPPI enjoys at least linear convergence rates when the optimization is quadratic, which covers time-varying LQR systems. We then extend to more general nonlinear systems. Our theoretical analysis directly leads to a novel sampling-based MPC algorithm, CoVariance-Optimal MPC (CoVo-MPC) that optimally schedules the sampling covariance to optimize the convergence rate. Empirically, CoVo-MPC significantly outperforms standard MPPI by 43-54% in both simulations and real-world quadrotor agile control tasks. Videos and Appendices are available at \url{https://lecar-lab.github.io/CoVO-MPC/}.