MotionLM: Multi-Agent Motion Forecasting as Language Modeling

TL;DR

MotionLM models multi-agent motion prediction as a language modeling task, enabling joint, multimodal trajectory forecasting without explicit latent variables, achieving state-of-the-art results on the Waymo dataset.

Contribution

It introduces a novel language modeling approach for multi-agent motion prediction that simplifies training and inference, and improves accuracy over prior methods.

Findings

Achieves state-of-the-art performance on Waymo dataset

Produces joint distributions over agent futures in a single decoding step

Does not require explicit latent variables or post-hoc heuristics

Abstract

Reliable forecasting of the future behavior of road agents is a critical component to safe planning in autonomous vehicles. Here, we represent continuous trajectories as sequences of discrete motion tokens and cast multi-agent motion prediction as a language modeling task over this domain. Our model, MotionLM, provides several advantages: First, it does not require anchors or explicit latent variable optimization to learn multimodal distributions. Instead, we leverage a single standard language modeling objective, maximizing the average log probability over sequence tokens. Second, our approach bypasses post-hoc interaction heuristics where individual agent trajectory generation is conducted prior to interactive scoring. Instead, MotionLM produces joint distributions over interactive agent futures in a single autoregressive decoding process. In addition, the model's sequential factorization enables temporally causal conditional rollouts. The proposed approach establishes new state-of-the-art performance for multi-agent motion prediction on the Waymo Open Motion Dataset, ranking 1st on the interactive challenge leaderboard.