United States Department of Transportation
i
Real-Time Safety Diagnosis System for Connected Vehicles With Parallel Computing Architecture
-
2023-12-11
-
[PDF-3.20 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:05/01/2022 to 12/11/2023
-
Corporate Publisher:
-
Abstract:The primary aim of this project is to enhance our system from the previous STRIDE F4 project to a parallel computing version. The original F4 system, designated as Automatic Safety Diagnosis in Connected Vehicle (CV) Environment, established a computational pipeline for diagnosing near-crash events exclusively using Basic Safety Messages (BSMs). It was implemented using a sequential computing paradigm. The O6 project was conceived to expedite the system by transitioning it to a parallel version. The F4 system comprised a driving anomaly detection model (DAD), a conflict identification model (CIM), and the data-path connecting them. The DAD was primarily situated in the core cloud, while the CIM was positioned within the CVs. Throughout the O6 process, notable advancements in in-vehicle computers (IVCs) were uncovered. In order to align our system with real-world operations, we opted to fully migrate the DAD component to the IVCs. Recognizing Domain-Specific Design (DSD) as the future of parallel computing, we propose configuring DSD for IVCs based on three levels of abstractions: selecting the appropriate chip architecture, programming language, and parallelism module. For the CIM of our system, we recommend utilizing ARM architecture, the C programming language, and leveraging the built-in parallelism of the ARM chip. As for the DAD, we advocate for a complete migration to IVC, utilizing ARM architecture, the Python language on the CPU, and employing multiprocessing for parallel computing.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
