AgiBot releases embodied Agent framework Genie Master; Genie G2 is already on duty in convenience stores
AgiBot has released the embodied Agent framework Genie Master, which supports autonomous improvement after robot deployment and proposes the Embodied Execution Protocol (EEP), unifying four types of interfaces: capability, guidance, feedback, and continuity. The Genie G2 equipped with this framework is already on duty in a real convenience store, autonomously identifying the causes of failures and adjusting strategies in tasks such as packaged hot food preparation and picking and restocking. AgiBot will launch a 24-hour uninterrupted livestream on November 2, with 6 Genie G2 robots working continuously across four shifts.
Why read this
AI score 62
The original article presents the interface design of the embodied Agent framework and real-world convenience store operation cases, offering insight into the path of autonomous improvement after robot deployment.
Topics
Body · AI translation
Related topics top start Related topics top end Futures promotion begin Futures promotion end Miaopai begin Miaopai end Video player start Video player end Market chart begin Market chart end
Sina Tech News, October 27 afternoon: Zhiyuan today released the embodied Agent framework Genie Master. It is reported that the framework supports autonomous improvement after robot deployment, can continuously enhance capabilities in real operations, lays the foundation for cross-scenario replication and large-scale application, breaks out of simulation environments and out of the laboratory, and can validate Physical RSI (embodied recursive self-improvement) in real business scenarios.
Genie Master takes connecting judgment, execution, and feedback as its design starting point, and proposes the Embodied Execution Protocol EEP, unifying four types of interfaces: capability, guidance, feedback, and continuity. Based on this framework, after the Agent issues execution guidance in combination with on-site observations, information on the success or failure and blocking points after the robot executes can be returned in real time; when a task is interrupted, the system can retain completed progress, allowing the Agent to replan from the breakpoint, while also supporting dynamic scheduling of multiple machines.
More importantly, the robot can also autonomously trace back the causes of failure, adjust execution strategies, and complete operations by supplementing the viewing angle through a wrist camera, making execution results the basis for the next round of judgment.
At present, the Elf G2 equipped with this framework has already started work in a real convenience store. Without relying on manual demonstration and debugging, it autonomously locates the causes of failure, adjusts strategies, and accumulates experience, achieving continuous improvement in multiple complex and delicate tasks such as cooked food packaging and picking and restocking. On November 2, Zhiyuan will launch a 24-hour uninterrupted livestream, and 6 Elf G2 robots will undergo continuous operations across four shifts to complete their "probation-to-regular" assessment. news_keyword_pub,stock,sz300024 iframe below main text iframe below main text General statement General statement QR code at end start QR code at end start
QR code at end start QR code at end start Editor name and work code
Editor name and work code