Perception Methods for Adverse Weather Based on Vehicle Infrastructure Cooperation System: A Review.

Area of Science:

• Autonomous Driving Technology
• Environmental Perception Systems
• Intelligent Transportation Systems

Background:

• Autonomous driving relies heavily on environment perception.
• Adverse weather and sensor limitations reduce perception accuracy and field of view for intelligent connected vehicles (ICVs).
• Current ICVs are restricted to specific operational scenarios due to perception challenges.

Purpose of the Study:

• To analyze existing research on image/point cloud processing and cooperative perception for ICVs.
• To summarize key aspects including data pre-processing, multi-sensor data fusion, and vehicle-infrastructure cooperative perception.
• To identify future research directions for enhancing cooperative perception in adverse weather.

Main Methods:

• Review of data pre-processing techniques for point cloud and image data under snow, rain, and fog conditions.
• Analysis of multi-sensor data fusion studies, including image, point cloud, and combined image-point cloud fusion.
• Examination of vehicle-infrastructure cooperative perception strategies for information fusion and sharing.

Main Results:

• Data pre-processing methods effectively handle adverse weather challenges for sensor data.
• Multi-sensor fusion techniques improve perception robustness by combining complementary data.
• Cooperative perception strategies, including fusion and sharing, expand ICV perception range and optimize information distribution.

Conclusions:

• Cooperative perception is essential for overcoming limitations in autonomous driving, especially in adverse weather.
• Further research is needed to optimize data fusion and information sharing for robust ICV perception.
• Developing advanced cooperative perception methods will enable autonomous driving in more challenging environmental conditions.