Edge Network-Assisted Real-Time Object Detection Framework for Autonomous Driving

Autonomous vehicles (AVs) can achieve the desired results within a short\nduration by offloading tasks even requiring high computational power (e.g.,\nobject detection (OD)) to edge clouds. However, although edge clouds are\nexploited, real-time OD cannot always be guaranteed due to dynamic channel\nquality. To mitigate this problem, we propose an edge network-assisted\nreal-time OD framework~(EODF). In an EODF, AVs extract the region of\ninterests~(RoIs) of the captured image when the channel quality is not\nsufficiently good for supporting real-time OD. Then, AVs compress the image\ndata on the basis of the RoIs and transmit the compressed one to the edge\ncloud. In so doing, real-time OD can be achieved owing to the reduced\ntransmission latency. To verify the feasibility of our framework, we evaluate\nthe probability that the results of OD are not received within the inter-frame\nduration (i.e., outage probability) and their accuracy. From the evaluation, we\ndemonstrate that the proposed EODF provides the results to AVs in real-time and\nachieves satisfactory accuracy.\n