Real-Time Vehicle Motion Detection and Motion Altering for Connected Vehicle: Algorithm Design and Practical...

Area of Science:

• Automotive Engineering
• Sensor Fusion
• Machine Learning for Transportation

Background:

• Real-time vehicle motion detection is crucial for connected vehicles and safe driving.
• Existing systems may lack comprehensive real-time state data capture from multiple sensors.

Purpose of the Study:

• To develop a novel vehicle motion detection system (VMDS) for real-time detection of lane-change, turning, acceleration, and deceleration.
• To enhance driving safety by alerting drivers to potential crash risks, particularly distracted drivers.

Main Methods:

• Utilized mobile sensors, including Global Positioning System (GPS) and inertial measurement units (IMUs).
• Developed a dynamic time warping algorithm combined with principal component analysis (PCA) for data processing.
• Trained and evaluated the algorithm on an Android platform using urban and highway driving data.

Main Results:

• VMDS achieved over 76% average precision in detecting lane-changes and turns.
• VMDS achieved over 91% average precision in detecting speed, acceleration, and braking.
• Alerting tests demonstrated a gain of nearly two seconds for drivers to perform safe operations.

Conclusions:

• The developed VMDS effectively detects various vehicle motions in real-time.
• The system significantly improves driving safety by reducing distracted driving and providing timely alerts.
• The VMDS shows promise for enhancing the safety of connected vehicles.