BC-MPPI: A Probabilistic Constraint Layer for Safe Model-Predictive Path-Integral Control

2025-09-30

AI Summary

MPPI control is computationally efficient and gradient-free but lacks hard guarantees on state/input constraints. To address this, we propose BC-MPPI: a lightweight safety-enhancement layer that integrates an offline-trained Bayesian probabilistic constraint surrogate model. This surrogate generates interpretable scalar safety scores for each candidate trajectory, enabling automatic weighting and real-time probabilistic constraint inference. The surrogate supports versioned deployment, embeds seamlessly into MPPI without manual tuning or sample rejection, and preserves computational efficiency. Evaluated on a MuJoCo quadrotor platform, BC-MPPI strictly enforces user-specified violation probability thresholds in both static and dynamic obstacle environments while maintaining robust safety margins. These results demonstrate its suitability for certification workflows in verifiable autonomous systems.

Technology Category

• Natural Language Processing: Safety and Robustness
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

Model Predictive Path Integral (MPPI) control has recently emerged as a fast, gradient-free alternative to model-predictive control in highly non-linear robotic tasks, yet it offers no hard guarantees on constraint satisfaction. We introduce Bayesian-Constraints MPPI (BC-MPPI), a lightweight safety layer that attaches a probabilistic surrogate to every state and input constraint. At each re-planning step the surrogate returns the probability that a candidate trajectory is feasible; this joint probability scales the weight given to a candidate, automatically down-weighting rollouts likely to collide or exceed limits and pushing the sampling distribution toward the safe subset; no hand-tuned penalty costs or explicit sample rejection required. We train the surrogate from 1000 offline simulations and deploy the controller on a quadrotor in MuJoCo with both static and moving obstacles. Across K in [100,1500] rollouts BC-MPPI preserves safety margins while satisfying the prescribed probability of violation. Because the surrogate is a stand-alone, version-controlled artefact and the runtime safety score is a single scalar, the approach integrates naturally with verification-and-validation pipelines for certifiable autonomous systems.

Problem

• Ensures safe trajectory planning under probabilistic constraints
• Eliminates need for manual penalty tuning in robotic control
• Provides certifiable safety guarantees for autonomous systems

Innovation

• Attaches probabilistic surrogate to constraints
• Trains surrogate from offline simulation data
• Uses joint probability to scale trajectory weights