物理人工智能：推动工业运营的新时代（英）

W H I T EP A P E R Contents Foreword Executive summary Introduction 1What’s new: Breakthroughs in intelligent robotics 1.1Technological breakthroughs redefining robotic capabilities1.2Enhanced capabilities enabling end-to-end automation 2Where it is working: Frontier applications 2.1Revolutionizing the manufacturing value chain2.2Spotlight on the pioneers – transformation journeys 3How it scales: Technology platforms and partnerships 3.1The new physical AI technology stack3.2Strategic partnerships are essential 4Who leads it: Empowering the new industrial workforce 4.1A target picture for robotics and workforce development4.2A shift in skills and roles Conclusion: Time for action Contributors Endnotes Disclaimer This document is published by theWorld Economic Forum as a contributionto a project, insight area or interaction.The findings, interpretations andconclusions expressed herein are a resultof a collaborative process facilitated and ©2025 World Economic Forum. All rightsreserved. No part of this publication maybe reproduced or transmitted in any formor by any means, including photocopying Foreword Daniel KuepperManaging Director andSenior Partner, Boston Kiva AllgoodManaging Director, Amid mounting global pressures – from economicvolatility and geopolitical disruption to growingsupply-chain complexity, and labour and talentshortages – industrial operations are entering atransformative new phase. While these challengesare not new, heightened uncertainty has significantly Economic Forum’s Next Frontier of Operationsinitiative in collaboration with Boston ConsultingGroup – builds on a tradition of strategic foresight The insights presented here draw on the collectiveexperience of global manufacturers, roboticsinnovators and leading academic experts.Grounded in real-world use cases and, more At this inflection point, a new era of industrialautomation is emerging – powered by physicalAI. These intelligent robotic systems combineperception, reasoning and action, enabling a levelof autonomy and adaptability that marks a criticaljuncture in industrial automation. By bridging the But this transformation is not solely abouttechnology. It also requires the industrial workforceto be equipped with new skills to collaborate withintelligent systems and take on emerging roles. Weinvite all stakeholders – including manufacturers,policy-makers, researchers and technologists – to As physical AI becomes increasingly viable andstrategically essential, industry leaders are seekinga deeper understanding of how to make useof these innovations for sustainable, long-term Executive summary Technological breakthroughs are pushingthe boundaries of automation – tasks thatwere once too variable or cost-prohibitiveto automate are now both technically Automation is expanding opportunities across theentire industrial value chain. Early adopters arealready achieving significant results. For example,Amazon, operating the world’s largest roboticsfleet, has demonstrated how the integration ofmobile robots, AI-based sortation and generativeAI-guided manipulators can improve fulfilmentcentre performance. By orchestrating theseautonomous systems, next-generation facilitieshave realized 25% faster delivery, 30% more skilled Although traditional industrial robots arefoundational to automation, they have long beenconstrained by limited adaptability and highintegration costs. Today, the world is entering anew age of robotics defined by intelligence and Approaches such as training methods(reinforcement learning, imitation learning) andmultimodal foundation models1for robotics, aswell as dexterous hardware components (e.g. softgrippers, tactile sensors) are enabling robots tohandle variability, reason in context and adapt inreal time. Simplified deployment, such as through However, realizing such outcomes at scaledemands more than cutting-edge technology.It requires a future-ready automation Such advances lead to three foundational roboticssystems that will coexist in the future of industrialoperations, together forming a layered automationstrategy. These systems are complementary, each –Embedding the emerging AI technology stackinto the existing industrial toolchain and forgingecosystem partnerships across robotics, AI –Rule-based robotics, delivering unmatchedspeed and precision in structured, repetitive –Workforce transformation through reskillingand upskilling to enable human–machinecollaboration, and prepare workers for emerging –Training-based robotics, mastering variabletasks via reinforcement learning or imitation Manufacturers who act now and embed roboticsas a strategic asset will lead the next phase ofindustrial competitiveness – shaping a futurein which intelligent automation becomes a –Context-based robotics, capable of zero-shot learning2and execution in unpredictableprocesses and new environments (e.g. robot Introduction Manufacturers must embrace intelligent No longerconfined toisolated eff