Special Issue on next-generation computing platforms for AI-enabled autonomous driving (SI:AIAD18)
摘要截稿:
全文截稿: 2019-03-15
影响因子: 2.552
期刊难度:
CCF分类: B类
中科院JCR分区:
• 大类 : 计算机科学 - 2区
• 小类 : 计算机:硬件 - 2区
• 小类 : 计算机:软件工程 - 2区
Overview
Autonomous Driving (AD) is expected to cause fundamental transformations to the transportation industry, with the prospect of significantly improving safety and comfort of the passengers, and reducing congestion of the transportation infrastructure. Artificial Intelligence/Machine Learning (AI/ML) is the key enabling technology for AD, impacting the entire processing pipeline, including perception and sensor fusion, path and trajectory planning, and low-level control. Specifically, Deep Learning and Deep Reinforcement Learning are the most effective AI/ML techniques used in AD today. While model training are very computation-intensive and are typically performed offline and in the cloud, runtime model inference also demand significant computing power from the in-vehicle embedded computing platform, which may be a heterogeneous mixture of multicore CPUs, GPUs, DSPs, FPGAs and ASICs. A number of vendors have provided commercial offerings of high-performance computing platforms designed for autonomous driving, e.g., NVidia’s DRIVE PX2, and others. Design and implementation of applications on top of such platforms bring many challenging research issues of performance, efficiency, power-consumption, reliability, dependability, security, and so on. This special section aims to present a collection of papers on the following topics in the context of computing/communication systems for AD: