Lessons Learned from Accident of Autonomous Vehicle Testing: An Edge Learning-aided Offloading Framework

TL;DR

This paper introduces an edge learning-based offloading framework for autonomous vehicles that enhances inference accuracy and reduces latency by optimizing task offloading to edge servers, considering wireless communication and computing constraints.

Contribution

It presents a novel optimization framework for offloading deep learning tasks in autonomous driving, balancing accuracy and latency under wireless and computational constraints.

Findings

Simulation results show improved inference accuracy.

The framework effectively balances latency and accuracy.

Superiority over existing offloading methods is demonstrated.

Abstract

This letter proposes an edge learning-based offloading framework for autonomous driving, where the deep learning tasks can be offloaded to the edge server to improve the inference accuracy while meeting the latency constraint. Since the delay and the inference accuracy are incurred by wireless communications and computing, an optimization problem is formulated to maximize the inference accuracy subject to the offloading probability, the pre-braking probability, and data quality. Simulations demonstrate the superiority of the proposed offloading framework.