System Overview
The AITRIOS's Platform
The AITRIOS's platform streamlines edge AI implementation without requiring specialized expertise across multiple domains.
The platform's key strength lies in its "mix and match" capability, allowing integration of services and applications from Sony, third party products, and custom-built solutions, supporting both cloud-native and on-premise options.
This significantly reduces development time, technical risk, and resource requirements while accelerating time-to-market for Edge AI solutions.
Edge Device
Capture, inference, and application logic
Raspberry Pi AI Camera
Sony AIH-IVRW2
Rayprus CSV26
Edge devices form the distributed intelligence network of the AITRIOS's ecosystem, capturing visual data and performing AI inference directly at the source, minimizing latency and bandwidth requirements.
AITRIOS's devices based on the groundbreaking IMX500 Intelligent Vision sensor feature a stacked sensor structure combining a CMOS sensor with a powerful DSP and dedicated on-chip SRAM to enable high-speed edge AI processing without needing a round trip to a server. This can reduce infrastructure costs, enhance efficiency, and accelerate edge AI deployment.
Core capabilities and advantages
Image sensors
High-quality image capture components optimized for AI processing
Inference engine
On-device AI processing capabilities that run models locally
Connectivity options
Multiple communication protocols (Wi-Fi, Ethernet) for flexible deployment
High efficiency system
A system designed to optimize bandwidth requirements and minimize power needs by leveraging the capabilities of the IMX500
Security features
Built-in mechanisms to protect data and prevent unauthorized access
Address privacy concerns
Help address personal data concerns effectively
*Outputting the processed image metadata only; the image data never leaves the chip, if you don't want it to
Console
Device management
Console is a central management hub of the platform, providing a unified interface to control all aspects of edge AI deployment, from device provisioning to application management.
Console can be hosted in the cloud or installed on a local machine.
The Console API allows device management and monitoring directly from other applications.
Edge AI management features
Device management
Tools for registering, monitoring, and managing edge devices at scale
Application deployment
Streamlined workflows for deploying AI models and applications to edge devices
Fleet management
Capabilities for organizing devices into logical groups and applying configurations at scale
Version control
Systems to manage and track versions of deployed models and applications
Configuration
Tools to adjust device settings and parameters without physical access
Conversion Service
AI preparation
The Conversion Service bridges the gap between standard AI development frameworks and edge deployment, ensuring models are compiled for the specific edge devices.
Conversion Service can be accessed in the cloud through Console or installed on a local machine.
Conversion features
Model compilation
Compile your AI models to run on AITRIOS's edge devices
Packager
Package your model for specific edge device hardware, with support to produce encrypted or unencrypted packages
AI Services
Data preparation and retraining
AI Tools from Sony and other suppliers in the AITRIOS's marketplace provide comprehensive capabilities for the entire AI model lifecycle, from initial data preparation to continuous model improvement through retraining.
For example, Brain Builder for AITRIOS and Studio Templates let you create and export customized AI models - with no coding required.
AI development lifecycle tools
Data collection
Tools to gather and organize training data from multiple sources
Annotation workspace
Interfaces for efficient labeling and annotation of training data
Model selection
Library of pre-trained models and architectures optimized for edge deployment
Training pipeline
Automated workflows for model training with configurable parameters
Performance evaluation
Tools to assess model accuracy, precision, and computational efficiency
Continuous learning
Systems for iterative model improvement based on real-world performance data
Solution App
Insights and actions
AITRIOS provides sample solution apps to demonstrate how versatile applications can collect and transform the inference data, and then combine visualization, alert functions, and system integrations to create comprehensive business solutions. You can tailor solution apps to your own specific operational needs, with nearly infinite integration possibilities.
The capstone of your implementation, your solution app is often customized for your own application, and can be very highly specialized.
The solution app can collect and transform the inference data from AITRIOS into actionable intelligence that drives business decisions and operational improvements.
AITRIOS offers sample apps and their source code to get started with.
Solution app capabilities
API integration
Robust APIs allow seamless connection with existing business systems and third-party applications for storage, analysis, and reporting
Data analytics
Apply advanced analytics to identify trends and patterns across collected inference data
Dashboards
Use data visualization libraries, or build custom web apps or mobile apps
Alert management
Implement notification systems for critical events identified by edge AI processing
Workflow automation
Create tools to automate responses to specific AI detections or patterns