Steering Wheel Contact Detection Using Camera-Torque Confidence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing torque sensor systems in vehicle steering systems face challenges in accurately detecting operator contact with the steering wheel, especially when certain portions of the operator are not visible within the camera's field-of-view.
Innovation Solution
The system employs a camera mounted on the dashboard to capture images of the operator, constructs a nodal model, estimates obscured feature locations, and determines a confidence level regarding operator contact with the steering wheel. Based on this confidence level, the system either utilizes torque measurements from a sensor on the steering column or actuates a notification for the operator to reposition their hands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a torque sensor is used to detect operator contact with the steering wheel, then the steering system can measure torque input, but the system cannot reliably distinguish between operator contact and foreign objects on the steering wheel
Solution Approach 1:
The patent combines multiple detection methods: torque sensing and computer vision (camera-based operator monitoring). The system merges data from the torque sensor with visual information from a camera that captures images of the operator and steering wheel, allowing the system to cross-validate whether detected torque corresponds to actual operator contact or foreign objects.
Solution Approach 2:
The patent introduces an intermediary processing system that correlates torque sensor data with camera-based operator position analysis. This intermediary layer processes both data streams together, using the camera as a mediator to verify whether the torque signal originates from operator hands or foreign objects, thereby resolving the ambiguity in torque-based contact detection.
2Device complexity
If the camera field-of-view is limited, then the system can focus on specific operator regions, but certain portions of the operator become obscured and invisible
Solution Approach 1:
The patent addresses the limited camera field-of-view by transitioning from a single 2D camera view to a multi-dimensional analysis approach. The system uses multiple cameras positioned at different locations and angles, creating a multi-perspective 3D reconstruction of the operator and steering wheel, thereby eliminating blind spots while maintaining individual camera simplicity.
Solution Approach 2:
The patent segments the operator monitoring task across multiple camera systems, each responsible for capturing specific regions or angles. By dividing the overall detection field into multiple segments covered by different cameras, the system ensures complete coverage of the operator and steering wheel without requiring any single camera to have an excessively wide or complex field-of-view.
3Reliability
If the system requires high confidence level for operator contact detection, then false notifications are reduced, but the system may miss genuine operator contact situations
Solution Approach 1:
The patent implements a confidence threshold system where the requirement for operator contact detection is set at a partial level rather than absolute certainty. By using a threshold-based approach (e.g., 70-80% confidence), the system achieves sufficient reliability to reduce false positives while maintaining responsiveness to genuine operator contact, avoiding the need for excessive confidence that would cause missed detections.
Solution Approach 2:
The patent incorporates feedback mechanisms where the system continuously monitors detection confidence levels and adjusts its response accordingly. When confidence exceeds the threshold, the system confirms operator contact; when confidence is marginal, the system may request additional verification or maintain a neutral state, creating a dynamic feedback loop that balances reliability and responsiveness based on real-time confidence assessment.
Data Source
AI summary
A system can include a computer having a processor coupled to a memory, the memory storing instructions executable by the processor to capture an image of an operator positioned in a steering position. The instructions can additionally be to determine, based on parameters extracted from the captured image, a confidence level that output data from a torque sensor represents torque applied by an operator to a steering element. The instructions can additionally be to actuate a component in a vehicle based on the confidence level.