Efficient Fusion and Task Guided Embedding for End-to-end Autonomous Driving
This work addresses computational efficiency for onboard vehicular computers in autonomous driving, representing an incremental improvement over existing lightweight methods.
The paper tackles the challenge of high computational demands in autonomous driving neural networks by introducing EfficientFuser, which reduces parameters by 37.6% and computations by 8.7% compared to state-of-the-art lightweight methods while maintaining a driving score within 0.4%.
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.