Cognitive System · Understanding & Decision-Making
Use terrain, road-surface variation and vehicle state to understand context and task goals and coordinate vehicle-wide systems.
JDO / EMBODIED INTELLIGENCE TECHNOLOGY
Embodied AI Technology for Vehicles and Robots
JDO AI System, an embodied AI core for vehicles and robots.
Explore the AI SystemJDO AI System
Support vehicle-wide coordinated control by combining in-cabin, external, and vehicle-state inputs.
Execution System · Execution & Feedback
Receive the coordinated intent, execute it in real time across domains, and return status feedback to support further understanding and decisions.
- Driving
- Driving strategy
- Chassis
- Suspension and attitude
- Cockpit
- Interaction and prompts
One System, Multiple Platforms
A shared foundation for vehicles and robots, adapted to each platform
Layered Memory
Tasks, Experience & Long-Term Preferences
Adaptive Reasoning
Adjust reasoning depth to the task
Scenario Prediction
Anticipate what may happen next
Automotive Technology Experience
Provide the perception, understanding, and memory foundation for in-cabin interactions and tasks.
Vision-Language Understanding (VLM)
Connect vision and language to understand road conditions outside the cabin, objects, interfaces, and questions.
In-Vehicle Voice and On-Device Response
Enable natural requests in task workflows despite in-cabin noise, weak connectivity, or offline conditions.
State and Health-Related Sensing
When authorized and supported by available data, assist with assessments and reminders without replacing medical diagnosis.
Driver and Occupant Monitoring
Provide inputs on driver and passenger state, including child-related cues, to support reminders, interactions, and services.
Multimodal Fusion and Context Understanding
Connect vision, speech, language, and vehicle-state signals to form usable context.
Memory
Within the scope authorized by the user, retain preferences and task context so services can continue seamlessly.
Technology Stack Support
Models and algorithms, a runtime framework, task planning, and cloud integration work together to support applications built on the AI system.
JDO Canghai AI Core
Multimodal Models & Perception Algorithms
Connect users' words with visual context.
Natural Language
Use natural language models to support semantic understanding.
Vision-Language & Perception
Combine vision-language and perception algorithms to support image understanding and object recognition.
AIOS
Context Management & Agent Runtime
Use the runtime framework to adapt to different platforms.
Model inference efficiency is optimized for on-device deployment, taking chip computing capacity and operating conditions into account.
Runtime & Context
Manage task context and history, and orchestrate Agents.
Platform Integration & Adaptation
Use APIs, containers, and middleware to integrate the required models and capabilities.
Generative Components & Themes
Explore component generation, theme changes, and interface composition.
AI Agent
Intent Parsing, Task Planning & Tool Invocation
Contextual information and task objectives guide multiple agents as they work together to complete tasks in mobility services and robotics operations.
Intent & Requirements
Consolidate requirements and constraints across multi-turn conversations.
Planning & Invocation
Organize task steps, invoke tools and services as needed, and receive execution results.
GUI Understanding & Operation
Working with partner capabilities, explore interface understanding and goal-directed operation.
AI Cloud
API / MCP Integration & Model Service Management
Coordinate cloud resources and business service integration, collect operational feedback, and support continuous improvement.
Models & Knowledge Resources
Manage cloud-based models and knowledge resources.
Business Service Integration
Connect business services through APIs and MCP.
Service Operations
- Application Updates
- Runtime Monitoring