Acoustic Data Processor for Robot Ego-Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise reduction methods, such as spectral subtraction, are ineffective for automatic speech recognition in robots due to ego-noises generated by their own motion, especially in near-field scenarios, where far-field noise reduction solutions perform poorly.
Innovation Solution
An acoustic data processor that obtains a robot's motion status using angle, angular velocity, and angular acceleration data to retrieve and subtract templates from a database, effectively reducing noise through template subtraction or multi-channel noise reduction, depending on the motion status.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional noise reduction methods such as spectral subtraction are used, then general noise reduction is achieved, but they are ineffective for ego-noises generated by robot motion in near-field scenarios
Solution Approach 1:
The system pre-collects acoustic data during various robot motions and stores it in a database before actual speech recognition tasks. This preliminary data collection and template creation enables the system to have reference noise profiles ready in advance, making the noise reduction effective for ego-noises without requiring real-time adaptation.
Solution Approach 2:
The system changes the approach from generic spectral subtraction to motion-status-specific template subtraction. By categorizing noise reduction strategies according to different robot motion parameters (standing, walking, running, etc.), the system adapts the noise reduction method to match the specific ego-noise characteristics generated by each motion type.
2Reliability
If far-field noise reduction solutions are used, then distant noise is reduced, but performance drops considerably for near-field ego-noise sources
Solution Approach 1:
The system applies different noise reduction strategies tailored to specific noise sources and scenarios. Instead of using a uniform far-field approach, it implements motion-status-specific template subtraction that is locally optimized for near-field ego-noise characteristics, thereby improving performance for the specific harmful factor of robot-generated near-field noise.
3Measurement precision
If spectral subtraction is performed for each behavior, then noise reduction accuracy is improved, but it becomes practically unfeasible to handle various kinds of behaviors
Solution Approach 1:
The system creates a universal database structure that can handle multiple motion types (standing, walking, running, etc.) through a single integrated framework. The motion status classification system provides a unified approach that automatically selects the appropriate template based on current robot state, making the complex multi-behavior noise reduction practically feasible through automated classification and template selection.
Solution Approach 2:
The system automatically determines the current robot motion status and selects the corresponding template from the database without requiring manual intervention or complex real-time analysis. This self-service mechanism simplifies the operation while maintaining high noise reduction accuracy across various behaviors.
Data Source
AI summary
An acoustic data processor according to the present invention is used for processing acoustic data including signal sounds to reduce noises generated by a mechanical apparatus. The acoustic data processor includes a motion status obtaining section for obtaining motion status of the mechanical apparatus, an acoustic data obtaining section for obtaining acoustic data corresponding to the obtained motion status, and a database for storing various motion statuses of the mechanical apparatus in a unit time and corresponding acoustic data as templates. The acoustic data processor further includes a database searching section for searching the database to retrieve the template having the motion status closest to the obtained motion status; and a template subtraction section for subtracting the acoustic data of the template having the motion status closest to the obtained motion status from the obtained acoustic data to reduce noises generated by the mechanical apparatus.