The Level 3 driver takeover detection system primarily evaluates the driver's capability to undertake driving tasks, enabling an interactive takeover mechanism between the automated driving system and the driver. Employing an in-cabin camera and advanced deep learning algorithms, the system accurately detects head posture and gaze direction, even for drivers wearing glasses or masks, to assess driver attentiveness. Additionally, by analyzing eye closure states, the system promptly identifies signs of fatigue or drowsiness. When the system determines the driver is incapable of resuming control, it issues a warning, prompting the driver to take over or initiating a minimum-risk maneuver by the Level 3 decision-making controller to ensure driving safety.

1.Adapts to various lighting conditions, including day and night, and accommodates drivers wearing glasses or masks.

2.Head pose detection range: yaw angle: +60° to -60°, pitch angle: +20° to -20°.

3.Meets UN R157 driver takeover availability requirements. 

4.Alerts after the driver leaves the seat for more than 1 second. 

5.Camera resolution is 720P. 

6.Can operate on AI edge computing devices with a processing time of approximately 300ms (requiring approximately 10TOPS computing power).

This technology can be applied in various types of vehicles equipped with in-cabin monitoring system, such as car, bus, intercity bus, autonomous vehicle…etc.

By leveraging this technology, domestic companies developing driving monitoring systems can enhance their product development capabilities, expedite development timelines, reduce costs, and increase their competitiveness in the global market through advanced features such as AI-based facial detection and recognition, adaptive personalization calibration, gaze tracking, lightweight multi-task model detection, and driver availability assessment.