Adaptive Risk Sensitive Model Predictive Control with Stochastic Search
This work addresses risk-sensitive control for robotics and dynamical systems, offering an incremental improvement over existing methods.
The paper tackles optimizing Conditional Value-at-Risk for dynamical systems under uncertainty, using a stochastic search framework that outperforms a risk-sensitive distributional reinforcement learning method on pendulum and cartpole tasks, with simulation results on robotics applications.
We present a general framework for optimizing the Conditional Value-at-Risk for dynamical systems using stochastic search. The framework is capable of handling the uncertainty from the initial condition, stochastic dynamics, and uncertain parameters in the model. The algorithm is compared against a risk-sensitive distributional reinforcement learning framework and demonstrates outperformance on a pendulum and cartpole with stochastic dynamics. We also showcase the applicability of the framework to robotics as an adaptive risk-sensitive controller by optimizing with respect to the fully nonlinear belief provided by a particle filter on a pendulum, cartpole, and quadcopter in simulation.