In-Cabin Voice Recognition Using Occupant Intent and Body Cues
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice recognition systems in vehicles face challenges in accurately identifying the occupant, interpreting commands, and distinguishing wake-up commands due to background noise, leading to inefficiencies in hands-free operation.
Innovation Solution
A voice recognition system that utilizes a supervised multicategory machine learning algorithm to convert voice signals into tokens, combines eye and body tracking algorithms to determine occupant intent factors, and executes regression and pattern recognition to determine hands-free tasks based on image data, voice signals, and context, including mood and conversation history.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional voice recognition systems are used, then the system can operate with simple architecture, but the accuracy of identifying occupants and interpreting commands deteriorates due to background noise and lack of contextual understanding
Solution Approach 1:
The patent combines multiple data sources (voice signals, image data from cameras, sensor data) and multiple processing techniques (traditional voice recognition, machine learning algorithms, eye tracking, body tracking) into an integrated system. This merging of previously separate functions enables accurate occupant identification and command interpretation while resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The system performs multiple functions simultaneously: it identifies occupants, determines their mood, tracks eye and body movements, recognizes voice commands, and executes hands-free tasks. This multi-functionality approach allows the system to achieve high measurement precision across multiple parameters while managing complexity through unified processing.
2Productivity
If wake-up commands are required to activate the voice recognition system, then the system can operate with simple activation logic, but the productivity deteriorates due to the need for multiple utterances and interruption of continuous conversation
Solution Approach 1:
The system continuously monitors multiple input channels (voice, image, sensors) and automatically determines when the occupant intends to speak or interact, eliminating the need for manual wake-up commands. The system serves itself by detecting intent through eye tracking, body tracking, and voice signal analysis, enabling seamless continuous operation.
Solution Approach 2:
The system performs preliminary analysis of eye movements, body posture, and voice signals to detect intent before the occupant actually speaks. By identifying preparatory gestures or eye movements toward the vehicle's display or controls, the system activates voice recognition in advance, allowing continuous conversation without wake-up interruptions.
3Reliability
If only voice signals are used for command recognition, then the system can operate with simple input processing, but the reliability deteriorates due to background noise and inaccurate command interpretation
Solution Approach 1:
The system uses feedback from multiple sources to continuously refine command interpretation. Eye tracking provides feedback on where the occupant is looking, body tracking provides feedback on posture and gestures, and voice signal analysis provides feedback on spoken commands. This multi-channel feedback mechanism significantly improves reliability by cross-validating information and filtering out noise.
Solution Approach 2:
The system introduces intermediate processing layers including machine learning models that analyze raw data from cameras and sensors, and contextual analysis modules that interpret the relationship between different input channels. These intermediaries transform raw multi-channel data into reliable command interpretations, resolving the contradiction between simple processing and high reliability.
Data Source
AI summary
A voice recognition system includes one or more controllers that receive an electrical signal representative of a voice signal generated by an occupant and image data representative of a head and an upper body of the occupant. The controllers convert the electrical signal representative of the words spoken by the occupant into a sequence of tokens based on a supervised multicategory machine learning algorithm, generates one or more sentences based on the sequence of tokens, and executes one or more eye and body tracking algorithms to determine one or more occupant intent factors. The controllers determine an intended action of the occupant based on the occupant intent factors and a context of the voice signal generated by the occupant. The controllers determine a hands-free task based on the context of the voice signal generated by the occupant, the intended action, and the one or more sentences.