Identifying autonomous vehicles (AVs) (e.g., those with adaptive cruise control) from traffic stream benefits enhancing traffic safety, elevating roadway capacity, and assisting autonomous vehicle management. This study tests the feasibility of identifying AVs using externally observed vehicle trajectory information. Two learning-based models are utilized to conduct the identification with car-following trajectory information in a short time window as the input. Four car-following trajectory datasets involving AVs makes from different manufacturers are mixed to build a comprehensive identification model. Results show that AVs and human-driven vehicles (HVs) can be successfully identified with a very high accuracy, i.e., the long short-term memory network can correctly identify 98.17% of AVs and 94.14% of HVs. This verifies the feasibility of using existing infrastructure and economic technologies to identify AVs from HVs, which opens unprecedented data-driven opportunities to study and manage near-future mixed traffic.
Automated Vehicle Identification in Mixed Traffic
2021-09-19
387544 byte
Conference paper
Electronic Resource
English
Automated mixed traffic transit vehicle microprocessor controller
Tema Archive | 1981
|Automated mixed traffic vehicle design AMTV 2
NTRS | 1982
|Automated Mixed Traffic Vehicle Design AMTV 2
NTIS | 1982
|