DriveGPT: Scaling Autoregressive Behavior Models for Driving

2024-12-19

arXiv.org

AI Summary

This work addresses the challenge of scalable modeling for future driving state prediction in autonomous driving. We propose an autoregressive behavioral model based on large-scale Transformers, framing driving as a sequential decision-making task and performing autoregressive prediction of multi-agent future trajectories tokenized by state representations. To our knowledge, this is the first systematic study of the co-scaling laws among data volume, model parameter count, and computational resources. The model enables end-to-end joint optimization of planning and prediction. It is pre-trained on large-scale multimodal driving datasets and evaluated in closed-loop simulation. Experiments demonstrate robust closed-loop driving performance in complex real-world scenarios, significant improvements over state-of-the-art methods in prediction accuracy, and consistent performance gains with increasing data and model scale—empirically validating the critical benefits of data-driven scaling for behavioral modeling.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Computer Vision: Large Vision Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

We present DriveGPT, a scalable behavior model for autonomous driving. We model driving as a sequential decision-making task, and learn a transformer model to predict future agent states as tokens in an autoregressive fashion. We scale up our model parameters and training data by multiple orders of magnitude, enabling us to explore the scaling properties in terms of dataset size, model parameters, and compute. We evaluate DriveGPT across different scales in a planning task, through both quantitative metrics and qualitative examples, including closed-loop driving in complex real-world scenarios. In a separate prediction task, DriveGPT outperforms state-of-the-art baselines and exhibits improved performance by pretraining on a large-scale dataset, further validating the benefits of data scaling.

Problem

• Autonomous driving behavior modeling
• Sequential decision-making task
• Scalable transformer model prediction

Innovation

• Autoregressive transformer for driving
• Scaled model parameters significantly
• Pretraining on large-scale dataset