MTIL: Encoding Full History with Mamba for Temporal Imitation Learning

TL;DR

MTIL introduces a novel approach using State Space Models to encode entire observation histories efficiently, enabling long-horizon imitation learning in robotics beyond the limitations of traditional Markovian methods.

Contribution

This paper presents MTIL, a new framework combining World Model and Dynamical System concepts with linear-time recurrent dynamics for efficient, full-history encoding in imitation learning.

Findings

MTIL outperforms SOTA methods on simulated benchmarks.

MTIL effectively resolves long-term temporal ambiguities.

The approach is computationally feasible for high-dimensional, long-horizon tasks.

Abstract

Standard imitation learning (IL) methods have achieved considerable success in robotics, yet often rely on the Markov assumption, which falters in long-horizon tasks where history is crucial for resolving perceptual ambiguity. This limitation stems not only from a conceptual gap but also from a fundamental computational barrier: prevailing architectures like Transformers are often constrained by quadratic complexity, rendering the processing of long, high-dimensional observation sequences infeasible. To overcome this dual challenge, we introduce Mamba Temporal Imitation Learning (MTIL). Our approach represents a new paradigm for robotic learning, which we frame as a practical synthesis of World Model and Dynamical System concepts. By leveraging the linear-time recurrent dynamics of State Space Models (SSMs), MTIL learns an implicit, action-oriented world model that efficiently encodes the entire trajectory history into a compressed, evolving state. This allows the policy to be conditioned on a comprehensive temporal context, transcending the confines of Markovian approaches. Through extensive experiments on simulated benchmarks (ACT, Robomimic, LIBERO) and on challenging real-world tasks, MTIL demonstrates superior performance against SOTA methods like ACT and Diffusion Policy, particularly in resolving long-term temporal ambiguities. Our findings not only affirm the necessity of full temporal context but also validate MTIL as a powerful and a computationally feasible approach for learning long-horizon, non-Markovian behaviors from high-dimensional observations.