Efficient Fusion and Task Guided Embedding for End-to-end Autonomous Driving

TL;DR

EfficientFuser is a compact, efficient neural network architecture for autonomous driving that combines visual and sensor data with minimal computational resources, maintaining high safety and driving performance.

Contribution

The paper introduces EfficientFuser, a novel lightweight fusion model using EfficientViT and transformer decoders for end-to-end autonomous driving, reducing resource needs significantly.

Findings

Uses only 37.6% of parameters of state-of-the-art methods

Requires just 8.7% of the computational resources

Achieves near-top safety and driving scores

Abstract

To address the challenges of sensor fusion and safety risk prediction, contemporary closed-loop autonomous driving neural networks leveraging imitation learning typically require a substantial volume of parameters and computational resources to run neural networks. Given the constrained computational capacities of onboard vehicular computers, we introduce a compact yet potent solution named EfficientFuser. This approach employs EfficientViT for visual information extraction and integrates feature maps via cross attention. Subsequently, it utilizes a decoder-only transformer for the amalgamation of multiple features. For prediction purposes, learnable vectors are embedded as tokens to probe the association between the task and sensor features through attention. Evaluated on the CARLA simulation platform, EfficientFuser demonstrates remarkable efficiency, utilizing merely 37.6% of the parameters and 8.7% of the computations compared to the state-of-the-art lightweight method with only 0.4% lower driving score, and the safety score neared that of the leading safety-enhanced method, showcasing its efficacy and potential for practical deployment in autonomous driving systems.