홈 › Papers

When Does Predictive Inverse Dynamics Outperform Behavior Cloning?

2026-01-29 · Lukas Schäfer, Pallavi Choudhury, Abdelhak Lemkhenter, Chris Lovett, Somjit Nath, Luis França, Matheus Ribeiro Furtado de Mendonça, Alex Lamb, Riashat Islam, Siddhartha Sen, John Langford, Katja Hofmann, Sergio Valcarcel Macua arxiv

Behavior cloning (BC) is a practical offline imitation learning method, but it often fails when expert demonstrations are limited. Recent works have introduced a class of architectures named predictive inverse dynamics models (PIDMs) that combine a future-state predictor with an inverse dynamics model. While PIDMs often outperform BC, the reasons behind their benefits remain unclear. In this paper, we provide a theoretical explanation: PIDMs introduce a tradeoff. Conditioning the IDM on the predicted future state can significantly reduce variance, but the prediction itself introduces additional bias and variance. We establish conditions for PIDMs to achieve higher sample efficiency and lower prediction error than BC, with the gap widening when additional data sources are available. We validate the theoretical insights empirically in 2D navigation tasks, where BC requires up to five times (three times on average) more demonstrations than PIDM to reach comparable performance. Results are also illustrated in a complex 3D environment in a modern video game with high-dimensional visual inputs and stochastic transitions, where BC requires over 66\% more samples than PIDM.

📄 PDF Abstract BibTeX arXiv:2601.21718

Code (0)

등록된 구현이 없습니다.

Model-Based Inverse Reinforcement Learning from Visual Demonstrations

2020-10-18 · Neha Das, Sarah Bechtle, Todor Davchev, Dinesh Jayaraman 외

Scaling model-based inverse reinforcement learning (IRL) to real robotic manipulation tasks with unknown dynamics remains an open problem. The key challenges lie in learning good dynamics models, developing algorithms th…

model Model Predictive Control reinforcement-learning Reinforcement Learning+1

Derivative-free online learning of inverse dynamics models

2018-09-13 · Diego Romeres, Mattia Zorzi, Raffaello Camoriano, Silvio Traversaro 외

This paper discusses online algorithms for inverse dynamics modelling in robotics. Several model classes including rigid body dynamics (RBD) models, data-driven models and semiparametric models (which are a combination o…

Transfer from Simulation to Real World through Learning Deep Inverse Dynamics Model

2016-10-11 · Paul Christiano, Zain Shah, Igor Mordatch, Jonas Schneider 외

Developing control policies in simulation is often more practical and safer than directly running experiments in the real world. This applies to policies obtained from planning and optimization, and even more so to polic…

Friction Reinforcement Learning

Learning Coordinated Bimanual Manipulation Policies using State Diffusion and Inverse Dynamics Models

2025-03-30 · Haonan Chen, Jiaming Xu, Lily Sheng, Tianchen Ji 외

When performing tasks like laundry, humans naturally coordinate both hands to manipulate objects and anticipate how their actions will change the state of the clothes. However, achieving such coordination in robotics rem…

Imitation Learning Navigate

Learning to Spend: Model Predictive Control for Budgeting under Non-Stationary Returns

2026-04-29 · Nilavra Pathak, Smriti Shyamal, Prasant Mhasker, Christopher Swartz arxiv

We study finite-horizon budget allocation as a closed-loop economic control problem and evaluate receding-horizon Model Predictive Control (MPC) relative to reactive budgeting policies. Budgets are allocated periodically…