Gaze-Assisted Voice Control for Accurate In-Car Apparatus Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately specify a unique apparatus for operation based on driver voice recognition, leading to potential incorrect operations.
Innovation Solution
An information processing device that utilizes line-of-sight data to specify the apparatus to be operated, generating control commands based on voice data, and fallbacks to past line-of-sight data when voice data alone is insufficient.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If apparatus is specified based on voice recognition result, then operation convenience is improved, but specification accuracy deteriorates when unique apparatus cannot be identified
Solution Approach 1:
The patent introduces line-of-sight information as an intermediary element to mediate between voice recognition and apparatus specification. When voice recognition alone cannot uniquely identify an apparatus, the system uses camera-based line-of-sight detection to determine which apparatus the driver is looking at, combining both modalities to achieve accurate specification while maintaining operational convenience
2Device complexity
If only voice data is used for apparatus specification, then system complexity is reduced, but reliability of operation deteriorates
Solution Approach 1:
The system performs preliminary line-of-sight detection and apparatus candidate identification before final apparatus specification. By pre-processing visual information to identify which apparatus the driver is looking at, the system prepares alternative specification methods in advance that can be activated when voice recognition alone is insufficient, thereby improving reliability without significantly increasing operational complexity
3Measurement precision
If line-of-sight data is integrated with voice recognition, then apparatus specification accuracy is improved, but device complexity increases
Solution Approach 1:
The patent implements a dynamic apparatus specification system that adaptively switches between voice-only mode and combined voice-plus-line-of-sight mode. The system evaluates whether voice recognition alone can uniquely identify an apparatus, and only activates the more complex line-of-sight integration when necessary, thereby achieving high specification accuracy while minimizing the impact on device complexity through conditional activation
Data Source
Figure 1
Figure 2
AI summary
In an information processing device and an information processing method, line-of-sight data pertaining to the direction of a user's line of sight is stored, and when voice data has been acquired, an apparatus to be operated is specified from a plurality of apparatuses based on the stored line-of-sight data, a control command directed at the apparatus to be operated is generated based on the voice data, and the apparatus to be operated is controlled.