Artificial Proprioceptive Reflex Warning Using EMG in Advanced Driving Assistance System.

Area of Science:

• * Neuroscience and Artificial Intelligence in Automotive Safety
• * Human-Computer Interaction for Driver Assistance

Background:

• * Lane-changing accidents frequently result from drivers disregarding surrounding traffic.
• * Current Advanced Driving Assistant Systems (ADAS) rely on external sensors, potentially missing driver intent.
• * Split-second decisions in driving necessitate proactive accident prevention strategies.

Purpose of the Study:

• * To predict a driver's intended action (left/right turn) before it occurs using electromyography (EMG) signals.
• * To integrate neural signal-based intention prediction with perception data from an autonomous driving system (ADS).
• * To develop an enhanced ADAS that warns drivers of potential hazards before action onset.

Main Methods:

• * Electromyography (EMG) signals were recorded and classified for left-turn and right-turn intentions.
• * Lane and object detection were performed using camera and Lidar data to perceive surrounding vehicles.
• * A fusion approach combined predicted driver intention with environmental perception for warning generation.

Main Results:

• * Successful classification of intended driving actions (left/right turns) from EMG data.
• * Demonstrated efficacy of integrating EMG-based intention prediction with sensor-based perception.
• * Validation of the proposed system through online and offline experiments in real-world driving scenarios.

Conclusions:

• * Neural signal prediction offers a novel layer of safety for ADAS beyond traditional sensors.
• * The proposed system can generate timely warnings by anticipating driver actions.
• * This approach has the potential to significantly reduce accidents caused by lane-changing maneuvers.