DM-MPPI: Datamodel for Efficient and Safe Model Path Integral Control
This work addresses efficiency and safety challenges in real-time control systems, such as robotics, by providing an incremental improvement to MPPI methods.
The authors tackled the problem of computational inefficiency and safety in Model Predictive Path Integral (MPPI) control by extending the Datamodels framework to enable real-time influence prediction for sample pruning and adaptive constraint handling, achieving up to a 5× reduction in samples while maintaining performance and improving constraint satisfaction.
We extend the Datamodels framework from supervised learning to Model Predictive Path Integral (MPPI) control. Whereas Datamodels estimate sample influence via regression on a fixed dataset, we instead learn to predict influence directly from sample cost features, enabling real-time estimation for newly generated samples without online regression. Our influence predictor is trained offline using influence coefficients computed via the Datamodel framework across diverse MPPI instances, and is then deployed online for efficient sample pruning and adaptive constraint handling. A single learned model simultaneously addresses efficiency and safety: low-influence samples are pruned to reduce computational cost, while monitoring the influence of constraint-violating samples enables adaptive penalty tuning. Experiments on path-tracking with obstacle avoidance demonstrate up to a $5\times$ reduction in the number of samples while maintaining control performance and improving constraint satisfaction.