Driver Notification Logic for Autonomous Control Handover Risk
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Drivers engaging in non-driving actions during autonomous driving may disrupt the transition to manual control, making it difficult to continue autonomous driving safely without clear understanding of which actions necessitate alerts.
Innovation Solution
A notification device that detects driver actions using facial recognition, calculates a confidence rate, and determines appropriate messages to inhibit control-ending actions by multiplying the confidence rate with probabilities, notifying the driver when the effectiveness level exceeds a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the autonomous driving system issues alerts for nondriving actions that may interrupt driver availability, then safety is improved, but it becomes difficult for drivers to understand in advance what actions would cause alerts
Solution Approach 1:
The notification device performs preliminary action by notifying the driver in advance about detected nondriving actions that may lead to control ending. The system detects actions like smartphone operation or face covering, then issues notifications before these actions result in autonomous driving termination, giving drivers time to correct their behavior and maintain understanding of safe driving expectations
Solution Approach 2:
The system implements feedback by providing real-time notifications to drivers about their detected actions. When the detection unit identifies nondriving actions through the driver monitor camera, the notification unit immediately informs the driver, creating a closed-loop feedback system that helps drivers understand what actions are problematic and how to adjust their behavior to maintain autonomous driving
2Loss of information
If the system notifies drivers of all detected nondriving actions, then driver awareness is improved, but notification frequency increases causing driver annoyance
Solution Approach 1:
The notification device applies local quality by providing selective notifications based on the specific type and severity of detected actions. Rather than uniform notification for all actions, the system tailors notifications to specific nondriving actions like smartphone operation or face covering, notifying only when actions reach thresholds that may affect autonomous driving safety, thus balancing awareness with driver comfort
3Reliability
If the autonomous driving system requires driver monitoring to ensure safety, then reliability is improved, but it limits the driver's ability to engage in nondriving actions
Solution Approach 1:
The system implements self-service by enabling drivers to maintain autonomous driving through their own corrective actions. When the notification unit alerts drivers about detected nondriving actions, drivers can independently correct their behavior (put down smartphones, uncover faces) to prevent control ending, thus maintaining both safety and driver freedom without requiring constant system intervention
Data Source
AI summary
A notification device detecting a predetermined action of a driver of a vehicle in which travel is controlled by an autonomous driving system from a driver image including a face region representing a face of the driver and outputting a confidence rate indicating a certainty in the detection, specifying a message for inhibiting a control ending action which possibly follows the detected action and corresponds to an ending condition of control by the autonomous driving system, and notifying the driver of the specified message through an output device when an effectiveness level is higher than a predetermined notification threshold, the effectiveness level calculated by multiplying the confidence rate by a first probability of the control ending action being inhibited when notifying the driver of the specified message or by a second probability of the control ending action being performed when not notifying the driver of the specified message.