NOVEL TEST SCENARIO GENERATION TECHNOLOGY FOR PERFORMANCE EVALUATION OF AUTOMATED VEHICLE
|
Shuang Li 1, Wei Li 1, Penghui Li 2,3, Ping Ma 1, Ming Yang 1 |
1Harbin Institute of Technology 2Beijing Jiaotong University 3China Automotive Engineering Research Institute Co., Ltd. |
|
|
|
|
ABSTRACT |
As one of the critical technologies for performance evaluation of automated vehicles, the test scenario generation
has been widespread concerned. In this paper, we propose a novel test scenario generation technology based on optimized Latin
Hypercube Sampling (OLHS) and Test Matrix method (TM), named HIS-MPSO, which is efficient to generate the test scenario
that consider the complexity, coverage, and potential relationships of factors. Based on naturalistic driving data, numerous carfollowing
scenarios are generated by HIS-MPSO. Then, an adaptive cruise control system (ACC) are evaluated in terms of the
tracking errors, comfort, and safety using the generated scenarios. Results show that compared with other existing OLHS
algorithms, the HIS-MPSO can better restore the relationships among test factors existed in realistic traffic scenarios. |
Key Words:
Test scenario generation, Optimized latin hypercube sampling, Test matrix, Naturalistic driving data, Automated vehicle
|
|