Automotive Assistant Sightline Filtering for Intent-Aware Speech
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern automotive assistants face ambiguity in determining whether an utterance is directed at them, especially in multi-occupant scenarios, leading to potential misinterpretation and unnecessary responses.
Innovation Solution
The implementation of a speech filter that uses sightline detection to determine if an occupant's gaze intersects a designated ocular zone, thereby deciding whether to disregard or respond to an utterance, incorporating a sightline detector and logic circuit to assess the probability of the utterance being addressed to the assistant.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the automotive assistant responds to all utterances without wake word activation, then the ease of operation is improved, but the reliability deteriorates due to misinterpretation of utterances not directed at the assistant
Solution Approach 1:
The patent introduces an ocular zone as an intermediary spatial parameter between the occupant and the automotive assistant. The sightline detection system acts as a mediator that determines whether the occupant is directing their attention to the assistant before processing utterances. This intermediary mechanism resolves the contradiction by maintaining ease of operation (no wake word needed) while improving reliability (filtering out unintended utterances through sightline verification).
2Reliability
If the automotive assistant uses sightline detection to determine intent, then the reliability is improved, but the device complexity increases due to additional sensors and processing
Solution Approach 1:
The patent applies multi-functionality by utilizing existing vehicle cameras for multiple purposes: their primary function for monitoring and the secondary function for sightline detection. The ocular zone is defined as a virtual parameter in existing camera coordinate systems, allowing the same hardware to perform both traditional vision tasks and new intent-detection tasks without adding dedicated sensors, thus improving reliability while minimizing increased device complexity.
3Ease of operation
If the automotive assistant defines a fixed ocular zone for each occupant, then the ease of operation is improved, but the adaptability deteriorates when occupant positions or orientations change
Solution Approach 1:
The patent implements dynamics by making the ocular zone a movable and adjustable parameter rather than a fixed definition. The ocular zone can be dynamically repositioned based on detected occupant location, orientation, and head pose. This dynamic adaptation allows the system to maintain ease of operation while improving adaptability to changing occupant positions and orientations throughout the vehicle cabin.
Data Source
AI summary
A method comprising causing an automotive assistant in a vehicle to disregard an utterance made by an occupant of that vehicle based on a sightline of the occupant.