Camera-in-the-loop based test scenario generation method for pedestrian collision avoidance system

Area of Science:

• Automotive Engineering
• Computer Vision
• Robotics

Background:

• Pedestrian Collision Avoidance Systems (PCAS) are crucial for intelligent vehicles (IV) to prevent accidents.
• Camera-based PCAS face challenges due to complex operating environments, necessitating rigorous testing.
• Traditional simulation methods lack the fidelity required for comprehensive PCAS evaluation.

Purpose of the Study:

• To develop and validate a Camera-in-the-Loop (CIL) test platform for camera-based PCAS.
• To propose an efficient test scenario generation method (GBC) for PCAS.
• To evaluate the impact of scenario parameters on PCAS performance using statistical analysis.

Main Methods:

• Construction and validation of a CIL test platform with emphasis on image quality and functional confidence.
• Development of a Greedy Based Combination (GBC) method for accelerated test scenario generation.
• Application of Chi-square and two-factor ANOVA for analyzing scenario parameter influence on PCAS.

Main Results:

• The CIL platform demonstrated credibility through validated image quality and functional confidence.
• The GBC method accelerated test scenario generation by 12 times compared to traversal testing.
• GBC achieved comparable accuracy to traversal testing in identifying critical PCAS collision scenarios.

Conclusions:

• The proposed CIL test platform and GBC method provide a credible and efficient solution for PCAS testing.
• The GBC method significantly enhances test efficiency without compromising the ability to identify critical failure modes.
• This approach is vital for the reliable deployment of camera-based PCAS in intelligent vehicles.