Event
Aetina demonstrates MCP architecture enabling AI agents to control robotic arms via natural language
1 reports · 1 sources
At COMPUTEX 2026, Aetina explained and demonstrated an edge AI architecture: an AI agent accesses robot control functions through the Model Context Protocol (MCP) as a standardized abstraction layer, converting an operator's natural-language instructions into robotic arm motions; the on-site demo showed a digital human using voice control to make a dexterous hand grasp a soda can. The architecture separates the "cognitive brain" from the "motor brain," running on edge devices such as Aetina's SuperEdge MGX server and DeviceEdge Jetson Thor platform; the entire process is completed at the edge, allowing the robot to process information locally and maintain control over its own motion without continuous dependence on a cloud connection. The grasping demo used a series of predefined, validated actions. Aetina's Tim Lin recommended a phased deployment approach, starting with selecting predefined robot actions that meet existing safety requirements.
Report timeline
- Aetina builds Physical AI edge computing architecture: using MCP to let AI agents drive robotic arms
Aetina uses AI agents plus the Model Context Protocol (MCP) to convert an operator's natural-language instructions into robotic arm actions, and at COMPUTEX 2026 demonstrated a digital human voice-controlling a dexterous hand to grasp a soda can.
Robotics 24/7 ·