iMotion-LLM: Instruction-Conditioned Trajectory Generation

TL;DR

iMotion-LLM is a novel large language model integrated with trajectory prediction modules that generates safe, feasible, and instruction-aligned driving trajectories based on textual commands, advancing interactive and interpretable autonomous driving systems.

Contribution

It introduces a new multimodal framework combining LLMs with trajectory prediction, along with two datasets for instruction-based trajectory generation in autonomous driving.

Findings

Achieves 84% accuracy in direction feasibility detection.

Achieves 96% accuracy in safety evaluation of instructions.

Demonstrates effective context-aware trajectory generation.

Abstract

We introduce iMotion-LLM, a large language model (LLM) integrated with trajectory prediction modules for interactive motion generation. Unlike conventional approaches, it generates feasible, safety-aligned trajectories based on textual instructions, enabling adaptable and context-aware driving behavior. It combines an encoder-decoder multimodal trajectory prediction model with a pre-trained LLM fine-tuned using LoRA, projecting scene features into the LLM input space and mapping special tokens to a trajectory decoder for text-based interaction and interpretable driving. To support this framework, we introduce two datasets: 1) InstructWaymo, an extension of the Waymo Open Motion Dataset with direction-based motion instructions, and 2) Open-Vocabulary InstructNuPlan, which features safety-aligned instruction-caption pairs and corresponding safe trajectory scenarios. Our experiments validate that instruction conditioning enables trajectory generation that follows the intended condition. iMotion-LLM demonstrates strong contextual comprehension, achieving 84% average accuracy in direction feasibility detection and 96% average accuracy in safety evaluation of open-vocabulary instructions. This work lays the foundation for text-guided motion generation in autonomous driving, supporting simulated data generation, model interpretability, and robust safety alignment testing for trajectory generation models. Our code, pre-trained model, and datasets are available at: https://vision-cair.github.io/iMotion-LLM/.