Work

Survey on Embodied AI for Industry: From the Perspective of Industrial Robots

Year: 2026

Type: article

Abstract: Industrial robots underlie modern manufacturing automation, yet conventional deterministic control based on fixed trajectories and offline programming struggles under high-mix and flexible production. Embodied artificial intelligence (EAI) offers a promising alternative by coupling perception, reasoning, and action within closed-loop physical interaction, enabling industrial robots to adapt behaviors online rather than execute predefined tasks. Yet, general-purpose EAI remains difficult to deploy in industrial environments due to stringent requirements on precision, real-time performance, reliability, and safety. These challenges have motivated increasing interest in embodied artificial intelligence for industry (EAI4I). This paper presents a systematic survey of EAI4I from an industrial robotics perspective. Specifically, we first analyze the quantified requirements of industrial robots enabled by EAI4I. Afterwards, recent research progress is reviewed, covering core technologies for single- and multi-robot systems, dedicated hardware platforms, high-fidelity simulators, task-specific datasets, representative industrial application scenarios, and critical deployment challenges. Finally, promising directions toward EAI4I are discussed.

Source: National Science Review

DOI: 10.1093/nsr/nwag505

Authors: Haibin Yu, Chunhe Song, Yinlong Zhang, Yiming Sun, Yong Sun, Chu Wang

Institution: Chinese Academy of Sciences

Language: English

Topic: Robot Manipulation and Learning

Subfield: Control and Systems Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,000