An Automotive LiDAR Performance Test Method in Dynamic Driving Conditions

Area of Science:

• Automotive Engineering
• Sensor Technology
• Robotics

Background:

• Light Detection and Ranging (LiDAR) sensors are crucial for autonomous driving and Advanced Driver Assistance Systems (ADAS).
• Ensuring LiDAR sensor reliability under extreme weather is vital for automotive safety systems.

Purpose of the Study:

• To develop and demonstrate a performance test method for automotive LiDAR sensors in dynamic test scenarios.
• To evaluate LiDAR sensor performance degradation under various environmental conditions.

Main Methods:

• A spatio-temporal point segmentation algorithm using unsupervised clustering was proposed to isolate LiDAR signals from moving targets.
• An automotive-grade LiDAR sensor was tested in four harsh environmental simulations mimicking real-world US road conditions.
• Four vehicle-level tests with dynamic scenarios were conducted.

Main Results:

• LiDAR sensor performance can be significantly impacted by environmental factors.
• Identified key factors include sunlight intensity, object reflectivity, and sensor cover contamination.
• The proposed segmentation algorithm effectively separates signals from dynamic reference targets.

Conclusions:

• The developed test method provides a robust evaluation of automotive LiDAR sensors in dynamic and adverse conditions.
• Understanding environmental impacts is critical for designing redundant and reliable automotive sensor systems.
• Further research should focus on mitigation strategies for LiDAR performance degradation.