Clustering-Embedded Model Predictive Path Integral Control: Avoiding Averaging-Induced Failure and Enabling Efficient Cluster Selection for Dynamic Obstacles

2026-07-07

AI Summary

This work addresses the limitations of standard Model Predictive Path Integral (MPPI) control in dense or dynamic obstacle environments, where trajectory averaging often leads to indecisive maneuvers, trajectory coupling, or collisions. The authors propose an enhanced MPPI control law that integrates high-fidelity pruning with a clustering mechanism, uniquely incorporating DBSCAN density-based clustering and geometric directional features derived from collision reference points into the MPPI framework to effectively disentangle feasible trajectory modes in non-convex settings. Tailored cluster selection strategies—minimizing cost for static scenes and opposing obstacle flow for dynamic ones—are introduced. Implemented with CUDA-accelerated parallel rollouts and evaluated on the Isaac Gym reinforcement learning platform, the method eliminates avoidance hesitation and prevents persistent coupling with moving obstacles in 2D simulations. Real-world experiments on a UR5e manipulator demonstrate a 48% reduction in task completion time and a 12% shorter end-effector path.

Abstract

With the widespread availability of parallel computing hardware, sampling-based motion planning methods such as Model Predictive Path Integral (MPPI) control have become increasingly powerful for complex nonlinear systems in non-smooth task spaces. However, the sampling and forward-simulation pipeline in MPPI suffers from averaging-induced failure in cluttered environments, where the importance-weighted update averages incompatible rollouts and leads to hesitation or even collision when an obstacle lies directly ahead. This paper proposes Clustering-Embedded MPPI (CE-MPPI), a framework that architecturally resolves the averaging-induced failures inherent in standard MPPI within non-convex environments. Rather than simply mitigating interference, CE-MPPI redefines the control law by integrating a high-fidelity pruning and clustering stage. By leveraging density-based spatial clustering of applications with noise (DBSCAN) alongside a novel geometric direction feature that is extracted from collision-derived reference points, the system isolates feasible trajectory modes from the noise of infeasible rollouts. This is paired with an intelligent selection logic that optimizes for minimum cost in static scenes while actively steering opposite to obstacle flux in dynamic environments. Experiments in 2-D JAX-accelerated simulations show that CE-MPPI alleviates obstacle-front hesitation and avoids persistent coupling with moving obstacles in dynamic scenes. In particular, real-world tests on a 6-DoF UR5e manipulator with CUDA-parallel rollouts in Isaac Gym achieve a 48\% reduction in time-to-goal and a 12\% shorter end-effector path.

Problem

• Model Predictive Path Integral Control
• Averaging-Induced Failure
• Dynamic Obstacles
• Non-convex Environments
• Trajectory Clustering

Innovation

• Clustering-Embedded MPPI
• averaging-induced failure
• DBSCAN
• geometric direction feature
• dynamic obstacle avoidance