PRIX: Learning to Plan from Raw Pixels for End-to-End Autonomous Driving

TL;DR

PRIX is an end-to-end autonomous driving model that uses only camera data, employing a novel transformer module to predict safe trajectories efficiently from raw pixels, matching state-of-the-art performance.

Contribution

It introduces PRIX, a camera-only planning architecture with a new CaRT module, achieving high performance without explicit BEV or LiDAR, and is suitable for real-world deployment.

Findings

PRIX matches state-of-the-art on NavSim and nuScenes benchmarks.

PRIX is more efficient in inference speed and model size.

PRIX operates solely on camera data without explicit BEV or LiDAR.

Abstract

While end-to-end autonomous driving models show promising results, their practical deployment is often hindered by large model sizes, a reliance on expensive LiDAR sensors and computationally intensive BEV feature representations. This limits their scalability, especially for mass-market vehicles equipped only with cameras. To address these challenges, we propose PRIX (Plan from Raw Pixels). Our novel and efficient end-to-end driving architecture operates using only camera data, without explicit BEV representation and forgoing the need for LiDAR. PRIX leverages a visual feature extractor coupled with a generative planning head to predict safe trajectories from raw pixel inputs directly. A core component of our architecture is the Context-aware Recalibration Transformer (CaRT), a novel module designed to effectively enhance multi-level visual features for more robust planning. We demonstrate through comprehensive experiments that PRIX achieves state-of-the-art performance on the NavSim and nuScenes benchmarks, matching the capabilities of larger, multimodal diffusion planners while being significantly more efficient in terms of inference speed and model size, making it a practical solution for real-world deployment. Our work is open-source and the code will be at https://maxiuw.github.io/prix.