Autonomous Driving Handover Control Using Gaze-Based Torque Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Level 3 autonomous vehicles incorrectly determine human driver intervention when the intervention is accidental, leading to premature deactivation of the autonomous driving system, increasing the risk of accidents.
Innovation Solution
The method dynamically adjusts steering wheel torque thresholds and intervention times based on the human driver's forward gaze state, distinguishing between intentional and accidental steering wheel manipulation by setting different thresholds and times for forward and non-forward gaze states, and activating a minimal risk maneuver mode to ensure safe handover of control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the autonomous vehicle uses a fixed threshold torque value and fixed steering intervention time for determining human driver intervention, then the determination process is simple, but it incorrectly determines accidental manipulation as intentional intervention, increasing accident risk
Solution Approach 1:
The patent applies dynamics by making the threshold torque value and steering intervention time dynamic rather than fixed. The control unit adjusts these parameters in real-time based on the driver's gaze direction detected by the gaze detector. When the driver is looking forward, stricter thresholds are applied; when looking away, more lenient thresholds are used. This dynamic adaptation resolves the contradiction by improving reliability through context-aware determination while managing complexity through automated gaze-based parameter adjustment.
Solution Approach 2:
The patent changes the parameters of the determination control function based on the driver's gaze state. The threshold torque value and steering intervention time are modified as parameters according to whether the driver is looking forward or away. This parameter change approach allows the system to adapt its sensitivity to steering input based on the driver's attention state, improving accuracy without requiring a completely complex new determination mechanism.
2Speed
If the autonomous vehicle quickly deactivates the autonomous driving system upon detecting steering wheel manipulation, then the response speed is fast, but it deactivates the system when the driver is not ready for manual driving, increasing accident risk
Solution Approach 1:
The patent applies preliminary action by using the gaze detector to assess the driver's readiness before completing the control handover. The system detects steering wheel manipulation and initiates a handover process, but the actual deactivation is delayed until the gaze detector confirms the driver is looking forward and ready to take control. This preliminary check ensures safety while maintaining relatively fast response, resolving the contradiction between speed and reliability.
Solution Approach 2:
The patent uses feedback from the gaze detector to modulate the handover process. The control unit continuously monitors the driver's gaze direction and uses this feedback to determine whether to proceed with or delay the deactivation of the autonomous driving system. This feedback mechanism ensures that the system only hands over control when the driver is actually ready, improving safety without significantly compromising response speed.
3Measurement precision
If the autonomous vehicle uses a single threshold for all steering situations, then the control logic is simple, but it cannot distinguish between intentional and accidental steering wheel manipulation
Solution Approach 1:
The patent applies local quality by using different threshold values for different local conditions - specifically, different gaze directions. When the driver is looking forward, one set of threshold parameters is applied; when looking away, another set is used. This local differentiation allows precise distinction between intentional and accidental manipulation in each context while keeping the overall mechanism relatively simple through automated gaze-based selection.
Solution Approach 2:
The patent introduces the gaze detector as an intermediary that mediates between the steering wheel input and the determination of intentional manipulation. Rather than directly comparing steering input to a single threshold, the system uses gaze direction as an intermediary parameter to select appropriate thresholds. This intermediary approach improves measurement precision by adding contextual information while managing complexity through a clear mediation layer.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for controlling autonomous driving in an autonomous vehicle includes: determining whether a human driver is in a forward gaze state under an autonomous driving mode, setting a first steering wheel torque threshold and a first torque holding time, based on a result of determining whether the human driver is in the forward gaze state, determining whether human driver intervention has occurred, based on the first steering wheel torque threshold and the first torque holding time, and switching the autonomous driving mode to a manual driving mode when the human driver intervention has occurred.