Steering Wheel Hold Detection for Autonomous Driving Takeover Intent

The patent segments the detection task into multiple independent sensor channels: steering wheel touch sensors detect contact, while separate cameras monitor driver hand position and posture. By dividing the detection system into multiple independent components, the patent can cross-validate signals to distinguish genuine takeover intentions from accidental touches, thereby improving reliability while filtering false positives.

The patent introduces an intermediary analysis layer that processes signals from multiple sensors before determining driver intention. This intermediary layer integrates data from steering wheel sensors, cameras, and other monitoring devices, using fusion algorithms to evaluate whether detected touches represent genuine takeover intentions or accidental contacts, thus resolving the contradiction between sensitivity and false positive rate.

The patent designs the camera system to serve multiple functions: it monitors driver hand position on the steering wheel, tracks driver posture, and observes overall driver behavior. This multi-functional approach allows a single camera system to provide inputs for multiple detection algorithms, reducing the need for separate specialized sensors for each function and thereby managing device complexity while maintaining high detection reliability.

The patent merges the detection functions of multiple sensors into a unified processing architecture. Instead of having separate independent systems for each detection task, the patent combines steering wheel sensors, cameras, and processing units into an integrated driver intention detection system that shares computational resources and data processing pathways, thus reducing overall system complexity while maintaining comprehensive monitoring capabilities.