Optical Neural Network Trigger Control for Low-Power Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional neural network devices experience increased power consumption and processing load, necessitating a solution to perform control with reduced energy usage.

Innovation Solution

A neural network device utilizing an optical neural network, including a light receiving portion and a control unit, operates with low energy consumption by detecting a target object based on light intensity distribution, allowing the device to start control operations only when a target is detected.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a neural network is used to perform processing and control, then the control accuracy and functionality are improved, but the processing load and power consumption increase

Engineering Contradiction:

Improvecontrol functionalityVSAvoidpower consumption

Core Design Contradiction:

Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The system divides the control task into two segments: (1) optical neural network processing performed by the optical device 100, and (2) conventional control processing performed by the control device 200. The optical neural network handles feature extraction and object detection, while the control device executes predetermined controls based on the optical processing results. This segmentation reduces the computational burden on the control device and lowers overall power consumption.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent replaces conventional electronic neural network processing with an optical neural network implementation. The optical device 100 performs neural network computations using optical components (lens, mirror, optical modulator) instead of electronic processors, achieving the same computational functionality with lower power consumption and higher processing speed.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Speed

If the optical neural network processes information continuously, then the detection responsiveness is improved, but the power consumption increases

Engineering Contradiction:

Improvedetection responsivenessVSAvoidpower consumption

Core Design Contradiction:

SpeedVSUse of energy by moving object

Solution Approach 1:

The optical neural network processes information periodically rather than continuously. The system captures optical signals at specific intervals, processes them through the optical neural network, and triggers control actions only when necessary. This periodic operation maintains detection responsiveness while significantly reducing power consumption compared to continuous processing.

Inventive Principle:

Principle #19Periodic action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The device achieves low power consumption and reduced processing load by initiating control only after detecting a target object, thereby conserving energy until a trigger is recognized.

Implementation Method 1

a light receiving portion that receives light through an optical neural network

Methodology Applied

Scientific EffectOptical neural network processing:

Implementation Method 2

a light receiving portion that receives light through an optical neural network, and a control unit that performs a predetermined control with detection of a predetermined target object based on a signal corresponding to light received by the light receiving portion

Methodology Applied

Scientific EffectPhotoelectric conversion: Photoelectric Effect

Data Source

PatentUS12615430B2Neural network device and control method

Publication Date: 2026.04.28 SONY GROUP CORP

AI summary

A neural network device includes a light receiving portion that receives light through an optical neural network, and a control unit that performs a predetermined control with detection of a predetermined target object based on a signal corresponding to light received by the light receiving portion as a trigger.