Abstract | ||
---|---|---|
Accurately predicting the dynamics of robotic systems is crucial for model-based control and reinforcement learning. The most common way to estimate dynamics is by fitting a one-step ahead prediction model and using it to recursively propagate the predicted state distribution over long horizons. Unfortunately, this approach is known to compound even small prediction errors, making long-term predictions inaccurate. In this paper, we propose a new parametrization to supervised learning on state-action data to stably predict at longer horizons -- that we call a trajectory-based model. This trajectory-based model takes an initial state, a future time index, and control parameters as inputs, and predicts the state at the future time. Our results in simulated and experimental robotic tasks show that our trajectory-based models yield significantly more accurate long term predictions, improved sample efficiency, and ability to predict task reward. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/CDC45484.2021.9683134 | CDC |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Nathan O. Lambert | 1 | 2 | 3.09 |
Albert Wilcox | 2 | 0 | 0.34 |
Howard Zhang | 3 | 0 | 0.34 |
Kristofer S. J. Pister | 4 | 1213 | 346.01 |
Roberto Calandra | 5 | 105 | 13.42 |