HOLO-MPPI: Multi-Scenario Motion Planning via Hierarchical Policy Optimization
Researchers have developed HOLO-MPPI, a new framework for multi-scenario motion planning in robotics. This approach combines high-level policy learning with low-level stochastic optimal control. The system learns a high-level policy offline to propose scenario-robust plans, which then parameterizes a real-time MPPI controller online to adapt to local disturbances. Evaluations in autonomous driving scenarios demonstrate that HOLO-MPPI outperforms existing MPPI and end-to-end reinforcement learning baselines while maintaining real-time performance. AI