AutoNavi Introduces Major Overhaul to ABot Embodied System with New Five Core Models

On July 23, Amap, which belongs to Alibaba Group, revealed that its ABot embodied system has undergone a comprehensive full-stack upgrade. As part of this update, the company officially introduced five new core models: ABot-N1, ABot-M0.5, ABot-ER, ABot-AgentOS, and ABot-C0. This advancement enabled seamless integration of perception, decision-making, execution, and memory functions within the system, greatly improving the robot’s capability to handle various general tasks independently. It represents a significant step forward in the development of highly integrated embodied intelligence technologies.
As the world’s first full-stack embodied technology system, ABot combines world models, base models, and an embedded Agent architecture to create a self-evolving closed-loop mechanism. This system is sustained through a mutually reinforcing cycle that incorporates simulation training, physical interaction, and memory scheduling.
The five upgraded models correspond directly to the key functional components of robots. ABot-N1 serves as the general navigation foundation, functioning akin to a robot’s “feet.” It utilizes a dual-system design with fast and slow modes to balance long-term logical reasoning with real-time control. ABot-M0.5 acts as the general operation foundation, similar to a robot’s “hands.” It separates movement from manipulation into separate action streams while integrating frame-level implicit actions to ensure precise coordination among the eyes, hands, and feet. ABot-ER functions as the embodied brain, while ABot-AgentOS serves as the execution center, together forming the higher-level system architecture. The brain enhances logical decision-making related to spatial relationships and tasks, whereas the execution center allows for unified adaptation across different robot types such as humanoid, quadrupedal, and wheeled designs, along with cross-task multimodal lifelong memory management. ABot-C0 acts as the motion control system, functioning like a “motor nerve,” by establishing a unified behavioral framework for quadrupedal robots that ensures accurate conversion of decisions into physical actions.
At present, embodied intelligence is at a pivotal moment, shifting from relying on individual technological breakthroughs to enabling deployment in complex real-world scenarios. By creating a full-stack collaborative architecture that encompasses the “hands, feet, brain, central nervous system, and motor nerves,” Amap has addressed previous limitations caused by fragmented robotic capabilities. This approach not only boosts the generalization and self-evolution abilities of mechanical systems in real-world environments but also provides a robust systematic framework for the widespread commercial use of general-purpose robots across various industries and service sectors.
Related article
Claude Opus 5.2 Night Gray Launches with Faster Response, Solving Laziness Issue
Opus 5.2 quietly launched this morning, prompting many developers to notice that the updated model, Claude Opus 5.2, has begun a limited rollout within Claude Code.Yesterday evening, X users observed that invoking Opus 5 in Claude Code yielded perfor
Fireworks AI Unveils FireRouter with Opus: Cuts Encoding Costs by 57% with Minimal Accuracy Drop
Fireworks AI has introduced FireRouter with Opus, the industry’s first cache-aware routing system tailored for the Claude Opus series. Now accessible via a serverless endpoint, this independent routing model underwent over a month of internal A/B tes
NVIDIA Unveils Nemotron-Labs-Audex-30B-A3B Unified Audio Intelligence Model
As multimodal large models evolve rapidly, audio processing capabilities are frequently compromised—many models improve audio understanding at the expense of text logic. To address this, NVIDIA researchers have introduced Nemotron-Labs-Audex-30B-A3B
Related Special Topic Recommendations
Comments (0)
0/500

On July 23, Amap, which belongs to Alibaba Group, revealed that its ABot embodied system has undergone a comprehensive full-stack upgrade. As part of this update, the company officially introduced five new core models: ABot-N1, ABot-M0.5, ABot-ER, ABot-AgentOS, and ABot-C0. This advancement enabled seamless integration of perception, decision-making, execution, and memory functions within the system, greatly improving the robot’s capability to handle various general tasks independently. It represents a significant step forward in the development of highly integrated embodied intelligence technologies.
As the world’s first full-stack embodied technology system, ABot combines world models, base models, and an embedded Agent architecture to create a self-evolving closed-loop mechanism. This system is sustained through a mutually reinforcing cycle that incorporates simulation training, physical interaction, and memory scheduling.
The five upgraded models correspond directly to the key functional components of robots. ABot-N1 serves as the general navigation foundation, functioning akin to a robot’s “feet.” It utilizes a dual-system design with fast and slow modes to balance long-term logical reasoning with real-time control. ABot-M0.5 acts as the general operation foundation, similar to a robot’s “hands.” It separates movement from manipulation into separate action streams while integrating frame-level implicit actions to ensure precise coordination among the eyes, hands, and feet. ABot-ER functions as the embodied brain, while ABot-AgentOS serves as the execution center, together forming the higher-level system architecture. The brain enhances logical decision-making related to spatial relationships and tasks, whereas the execution center allows for unified adaptation across different robot types such as humanoid, quadrupedal, and wheeled designs, along with cross-task multimodal lifelong memory management. ABot-C0 acts as the motion control system, functioning like a “motor nerve,” by establishing a unified behavioral framework for quadrupedal robots that ensures accurate conversion of decisions into physical actions.
At present, embodied intelligence is at a pivotal moment, shifting from relying on individual technological breakthroughs to enabling deployment in complex real-world scenarios. By creating a full-stack collaborative architecture that encompasses the “hands, feet, brain, central nervous system, and motor nerves,” Amap has addressed previous limitations caused by fragmented robotic capabilities. This approach not only boosts the generalization and self-evolution abilities of mechanical systems in real-world environments but also provides a robust systematic framework for the widespread commercial use of general-purpose robots across various industries and service sectors.
Claude Opus 5.2 Night Gray Launches with Faster Response, Solving Laziness Issue
Opus 5.2 quietly launched this morning, prompting many developers to notice that the updated model, Claude Opus 5.2, has begun a limited rollout within Claude Code.Yesterday evening, X users observed that invoking Opus 5 in Claude Code yielded perfor
Fireworks AI Unveils FireRouter with Opus: Cuts Encoding Costs by 57% with Minimal Accuracy Drop
Fireworks AI has introduced FireRouter with Opus, the industry’s first cache-aware routing system tailored for the Claude Opus series. Now accessible via a serverless endpoint, this independent routing model underwent over a month of internal A/B tes
NVIDIA Unveils Nemotron-Labs-Audex-30B-A3B Unified Audio Intelligence Model
As multimodal large models evolve rapidly, audio processing capabilities are frequently compromised—many models improve audio understanding at the expense of text logic. To address this, NVIDIA researchers have introduced Nemotron-Labs-Audex-30B-A3B





Home






