Pedestrian Detection in Blind Area and Motion Classification Based on Rush-Out Risk Using Micro-Doppler Radar

Area of Science:

• Robotics and Intelligent Systems
• Sensor Technology
• Computer Vision

Background:

• Current remote sensing struggles with detecting pedestrians in blind areas and their dynamic behaviors like sudden rushes.
• This limitation poses significant safety risks for intelligent vehicles and robots.

Purpose of the Study:

• To develop and validate a radar-based method for detecting pedestrians in blind spots.
• To classify the risk levels associated with different pedestrian rush-out behaviors detected in these blind areas.

Main Methods:

• A novel radar-based technique was developed for pedestrian detection.
• Experiments were conducted in diverse environments (outdoor cars, indoor walls) to simulate blind areas.
• Clustering methods were employed for classifying pedestrian rush-out behavior risks.

Main Results:

• The radar technique successfully detected pedestrian motion in blind areas across different experimental settings.
• The clustering method effectively classified various rush-out behaviors based on their associated risks.
• The system demonstrated robust performance in both indoor and outdoor blind-area scenarios.

Conclusions:

• Radar technology offers a viable solution for detecting pedestrians in challenging blind spots.
• The proposed method enhances pedestrian safety by enabling risk-based classification of rush-out behaviors.
• This research contributes to the advancement of perception systems for autonomous vehicles and robots.