# When Does Predictive Inverse Dynamics Outperform Behavior Cloning?

arXiv: https://arxiv.org/abs/2601.21718 · HTML: https://mlsanity.com/paper/2601.21718/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Imitation learning methods, one of 13 directions in Offline RL and imitation learning from data.
- Builds on 21 of its 38 references that are in the map.
- First posted on arXiv 29 Jan 2026.

## Placed in

- Offline RL and imitation learning from data / [Imitation learning methods](https://mlsanity.com/offlinerl/imitation-learning-methods/)

## Builds on (most cited first, 21 of 21)

- [Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models](https://mlsanity.com/paper/1805.12114/) (2018)
- [Learning Universal Policies via Text-Guided Video Generation](https://mlsanity.com/paper/2302.00111/) (2023)
- [An Algorithmic Perspective on Imitation Learning](https://mlsanity.com/paper/1811.06711/) (2018)
- [Implicit Behavioral Cloning](https://mlsanity.com/paper/2109.00137/) (2021)
- [DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning](https://mlsanity.com/paper/2411.04983/) (2024)
- [Navigation World Models](https://mlsanity.com/paper/2412.03572/) (2024)
- [Imitating Human Behaviour with Diffusion Models](https://mlsanity.com/paper/2301.10677/) (2023)
- [Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulation](https://mlsanity.com/paper/2412.15109/) (2024)
- [mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAs](https://mlsanity.com/paper/2512.15692/) (2025)
- [Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning](https://mlsanity.com/paper/2407.15007/) (2024)
- [Theia: Distilling Diverse Vision Foundation Models for Robot Learning](https://mlsanity.com/paper/2407.20179/) (2024)
- [Counter-Strike Deathmatch with Large-Scale Behavioural Cloning](https://mlsanity.com/paper/2104.04258/) (2021)
- [Guaranteed Discovery of Control-Endogenous Latent States with Multi-Step Inverse Models](https://mlsanity.com/paper/2207.08229/) (2022)
- [Latent Diffusion Planning for Imitation Learning](https://mlsanity.com/paper/2504.16925/) (2025)
- [Representation Learning with Multi-Step Inverse Kinematics: An Efficient and Optimal Approach to Rich-Observation RL](https://mlsanity.com/paper/2304.05889/) (2023)
- [Understanding the bias-variance tradeoff of Bregman divergences](https://mlsanity.com/paper/2202.04167/) (2022)
- [Multistep Inverse Is Not All You Need](https://mlsanity.com/paper/2403.11940/) (2024)
- [A Generalized Bias-Variance Decomposition for Bregman Divergences](https://mlsanity.com/paper/2511.08789/) (2025)
- [Bias-variance decompositions: the exclusive privilege of Bregman divergences](https://mlsanity.com/paper/2501.18581/) (2025)
- [Adapting a World Model for Trajectory Following in a 3D Game](https://mlsanity.com/paper/2504.12299/) (2025)
- [PcLast: Discovering Plannable Continuous Latent States](https://mlsanity.com/paper/2311.03534/) (2023)

## Built on by (most cited first, 3 of 3)

- [Latent Geometry Beyond Search: Amortizing Planning in World Models](https://mlsanity.com/paper/2605.08732/) (2026)
- [FF-JEPA: Long-Horizon Planning in World Models with Latent Planners](https://mlsanity.com/paper/2606.09311/) (2026)
- [Augmentations for Robust and Efficient Imitation Learning in Streamed Video Games](https://mlsanity.com/paper/2607.14200/) (2026)

## Abstract (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.