|
Work item:
|
F.CAV-MEMSF
|
|
Subject/title:
|
Metrics and evaluation methods for the robustness of multi-sensor fusion-based perception in connected and automated vehicles
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2027-01 (Medium priority)
|
|
Liaison:
|
ISO/TC 22/SC 31
|
|
Supporting members:
|
Ministry of Industry and Information Technology (MIIT); Chongqing Changan Automobile Co., Ltd.; China Information and Communication Technology Group Co., Ltd. (CICT)
|
|
Summary:
|
Multi-sensor fusion perception is vital for connected and automated vehicles to handle complex environments using data from sensors like cameras, LiDAR, and V2X communication. However, challenges such as extreme weather, sensor failures, and communication issues can reduce perception accuracy and stability. This recommendation addresses these challenges by defining robustness evaluation methods for MSF modules, focusing on tasks like collaborative localization, object detection, and lane detection. It also specifies evaluation metrics, including accuracy, precision, and IoU, to enhance system reliability and ensure CAV safety in dynamic driving scenarios.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-02-27 22:04:24
|
|
Last update:
2025-02-28 10:22:39
|
