Information Theoretic Model Predictive Control: Theory and Applications to Autonomous Driving
About
We present an information theoretic approach to stochastic optimal control problems that can be used to derive general sampling based optimization schemes. This new mathematical method is used to develop a sampling based model predictive control algorithm. We apply this information theoretic model predictive control (IT-MPC) scheme to the task of aggressive autonomous driving around a dirt test track, and compare its performance to a model predictive control version of the cross-entropy method.
Grady Williams, Paul Drews, Brian Goldfain, James M. Rehg, Evangelos A. Theodorou• 2017
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Quadruped Locomotion | Unitree Go2 Robot Simulation | Cost46.73 | 6 | |
| Trajectory Optimization | Hopper | Computational Time (s)26.4 | 5 | |
| Trajectory Optimization | Ant | Computational Time (s)16 | 5 | |
| Trajectory Optimization | Humanoid Running | Computational Time (s)29.6 | 5 | |
| Trajectory Optimization | Half Cheetah | Computational Time (s)26.7 | 5 | |
| Trajectory Optimization | Humanoid Standup | Computational Time (s)17.7 | 5 | |
| Trajectory Optimization | Push T | Time (s)1.03e+3 | 5 | |
| Trajectory Optimization | Walker2D | Computational Time (s)34.7 | 5 | |
| Trajectory Optimization | 8x8m Open Environment Goal (6, 0) | Success Rate100 | 4 | |
| Trajectory Optimization | 8x8m Cluttered Environment Goal (6, 0) | Success Rate (SR)100 | 4 |
Showing 10 of 13 rows