What is Wuneng
Wuneng is an embodied intelligence system launched by SenseTime for robots and smart devices. The platform is based on SenseTime's "Kaiwu" world model and combines multimodal large model technology to empower robots with powerful perception, decision-making, and action capabilities. The platform includes a perception layer (multisensor input such as vision, voice, and touch), a decision-making layer (LLM + world model collaborative planning), and an action layer (end-to-end control of robotic arms and mobile chassis). The platform can empower various terminal hardware such as robots, supports embedding into edge-side chips, adapts to multiple scenarios, can generate multi-view videos, build a 4D real world, and enable autonomous execution of complex tasks.
Wuneng's Main Features
- Perception Features：
- Multimodal Perception Fusion: Based on input from multiple sensors such as vision, voice, and touch, it achieves comprehensive perception and understanding of the environment.
- Environmental Perception: Recognizes objects, people, and spatial layouts in real time, providing accurate environmental information for decision-making.
- Decision-Making Features：
- Collaboration Between LLM and World Model: Combines large language models and world models to integrate language understanding with knowledge of the physical world and formulate reasonable action strategies.
- Task Planning: Generates detailed execution plans according to task objectives to ensure efficient task completion.
- Action Features：
- End-to-End Control: Precisely controls robotic arms and mobile chassis to achieve accurate execution of complex movements.
- Action Execution and Feedback: Makes real-time adjustments during task execution to ensure flexibility and reliability.
-
Scenario Adaptability: Supports embedding into edge-side chips and adapts to various smart devices and scenarios.
Wuneng's Technical Characteristics
-
Layered Design: The layered design of the perception layer, decision-making layer, and action layer makes the system structure clear and facilitates development and optimization.
-
Modularity: Each layer is internally based on a modular design, supporting flexible expansion and customized development.
-
Collaborative Work: The layers work closely together through efficient communication mechanisms to ensure the overall performance of the system.
-
Real-Time Capability: Supports real-time data processing and feedback to ensure that the system can respond quickly to environmental changes.
-
Safety: Designs multiple safety redundancies for critical functions to ensure that the system can still operate safely under abnormal conditions.
Wuneng's Application Scenarios
- Home Services: Assists with household tasks (such as sweeping and organizing) while also providing companionship and interaction, improving the quality of home life.
- Industrial Automation: Automatically transports goods in logistics warehouses, optimizes the assembly and quality inspection of parts on production lines, and improves efficiency.
- Autonomous Driving: Perceives road conditions in real time and provides autonomous driving assistance functions such as automatic parking and emergency avoidance, improving driving safety.
- Medical Care: Assists patients with rehabilitation training, monitors patient status in wards, and provides auxiliary support for medical staff.
- Commercial Services: Provide shopping advice to customers in the mall, assist with sales guidance, and simultaneously conduct real-time security monitoring to ensure the safety of the commercial premises.