AutoTraces: Autoregressive Trajectory Forecasting via Multimodal Large Language Models

TL;DR

AutoTraces leverages multimodal large language models with a novel trajectory tokenization scheme and chain-of-thought reasoning to improve long-term robot trajectory forecasting in human environments.

Contribution

It introduces a new trajectory tokenization method and automated reasoning mechanism, extending LLMs to physical coordinate spaces for enhanced trajectory prediction.

Findings

Achieves state-of-the-art accuracy in long-horizon forecasting

Demonstrates strong cross-scene generalization

Supports flexible-length trajectory predictions

Abstract

We present AutoTraces, an autoregressive vision-language-trajectory model for robot trajectory forecasting in humam-populated environments, which harnesses the inherent reasoning capabilities of large language models (LLMs) to model complex human behaviors. In contrast to prior works that rely solely on textual representations, our key innovation lies in a novel trajectory tokenization scheme, which represents waypoints with point tokens as categorical and positional markers while encoding waypoint numerical values as corresponding point embeddings, seamlessly integrated into the LLM's space through a lightweight encoder-decoder architecture. This design preserves the LLM's native autoregressive generation mechanism while extending it to physical coordinate spaces, facilitates modeling of long-term interactions in trajectory data. We further introduce an automated chain-of-thought (CoT) generation mechanism that leverages a multimodal LLM to infer spatio-temporal relationships from visual observations and trajectory data, eliminating reliance on manual annotation. Through a two-stage training strategy, our AutoTraces achieves SOTA forecasting accuracy, particularly in long-horizon prediction, while exhibiting strong cross-scene generalization and supporting flexible-length forecasting.