Model Predictive Path Integral Control: From Theory to Parallel Computation

By Grady Williams, Andrew Aldrich, Evangelos A. Theodorou. 2017

510 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering

View Paper: https://www.semanticscholar.org/paper/ddc1d2830c5979bb07ca631fa711e1e384d5ddfa

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

Citations in the corpus

Implicit Generation and Generalization with Energy Based Models (2018, methodology): “For MPPI, we use 30 steps of simulation with 5 noise simulations per step.”
Model-Based Offline Planning (2020, methodology): “…model-based controller can quickly learn to drive a vehicle on a dirt track, the BADGR robot [21] also uses Model-Predictive Path Integral (MPPI) [22] with a learned model to learn to navigate to novel locations, Yang et al. [23] show good results learning legged locomotion policies using MPC…”
Model Based Planning with Energy Based Models (2019, methodology): “To sample from the defined probability distribution, we use Model Predictive Path Integral (MPPI) algorithm [21], which is shown to converge to the full posterior distribution in [30].”
Information-Theoretic Model Predictive Control: Theory and Applications to Autonomous Driving (2017, methodology): “In contrast, prior work on model predictive path-integral control [26] is based on dynamic programming, and importance sampling is incorporated in a heuristic fashion.”
Reinforcement Learning: Theory and Algorithms (2019, background): “[Williams et al., 2017]), then it is often more effective to work with finite horizon, non-stationary models.”
Information theoretic MPC for model-based reinforcement learning (2017, methodology): “A more flexible MPC method is model predictive path integral (MPPI) control, a sampling-based algorithm which can optimize for general cost criteria, convex or not [12]– [14].”
VIP: Towards Universal Visual Reward and Representation via Value-Implicit Pre-Training (2022, methodology): “We use a publicly available implementation of MPPI 3 , and make no modiﬁcation to the algorithm or the default hyperparameters.”
Differentiable MPC for End-to-end Planning and Control (2018, background): “Okada et al. [2017] and Pereira et al. [2018] present a means to differentiate through path integral optimal control [Williams et al., 2016, 2017] and learn a planning policy end-to-end.”