Humanoid Robots & Embodied AI at World Robot Conference 2025: A Revolution in Service, Healthcare, and Logistics

Humanoid Robots & Embodied AI at World Robot Conference 2025: A Revolution in Service, Healthcare, and Logistics

Introduction

At the 10th World Robot Conference (WRC 2025) held in Beijing from August 8 to 12, 2025, more than 100 cutting-edge products and humanoid robot innovations were unveiled, capturing global attention. Hosted at the expansive Beiren Yichuang International Convention & Expo Center, the expo spanned over 52,000 sqm and featured 200+ domestic and international exhibitors showcasing embodied intelligence in healthcare, logistics, rescue, and beyond.

Under the theme “Making Robots Smarter, Making Embodied Agents More Intelligent”, WRC emphasized the transformative rise of robotic agents not just as tools, but collaborative teammates.

Embodied Intelligence on Display: From Healthcare to Logistics

Healthcare & Medical Robotics

Healthcare robotics emerged as a standout application. Robots equipped with AI and tactile sensors provided automated assistance for surgeries, patient care, and pharmacy services. Several models demonstrated capabilities such as precise medication dispensing and neural-impulse responsive prosthetics, illustrating a seamless blend of physical embodiment and smart decision-making.

Logistics & Service Robotics

Logistics and service applications made major strides. From agile humanoids capable of package handling to autonomous delivery systems, the conference featured robots that can navigate real-world environments. A highlight the newly opened Robot Mall 4S store in Beijing’s Yizhuang districtoffered a full-service experience for humanoid robots, including interactive demonstrations, diagnostics, and replacements. Featured robots included Tiangong, Unitree's G1, and UBTECH’s Walker S.

Competitive & Performance Robotics

Sporting and entertainment robotics offered a glimpse of embodied AI in dynamic action. The World Humanoid Robot Games, held August 15–17 alongside the conference, included soccer, sprints, long jump, gymnastics, dancing, and scenario-based challenges that simulate real-life tasks. Early previews like the 3-on-3 AI-controlled humanoid football matches demonstrated capabilities and limitations in agility and perception.

Core Technology: Embodied AI & Vision-Language-Action Models

The concept of embodied AI Integrating perception, language understanding, and motor action—was central to many models presented.

Helix, unveiled in February 2025, is a Vision-Language-Action (VLA) model for humanoid control. It uses a dual-system architecture: a VLM (System 2) for scene understanding and a visuomotor controller (System 1) for real-time action translation.

GR00T N1, released by NVIDIA in March, follows a similar dual-system VLA paradigm. Trained on heterogeneous data (robot trajectories, human video, synthetic), it achieves high performance in language-conditioned manipulation on the Fourier GR-1 platform.

Being-0, introduced in March 2025, integrates a foundation model with modular low-level skills via a “Connector” that translates planning into locomotion and manipulation. It enables long-horizon task execution with real-time efficiency on full-size humanoid platforms.

These embodied AI architectures are pushing the frontier of autonomous humanoid applications—from hospitals to warehouses.

China’s Robotics Leadership & Ecosystem

China continues to solidify its dominance in humanoid and industrial robotics: two-thirds of global robot patent applications, 556,000 industrial robots produced in 2024, and a \$10 billion+ robotics output centered in Beijing’s E-Town.

During WRC 2025, JD.com served as the exclusive global strategic partner, announced strategic plans to support robotics industry development and collaboration with names like Unitree and AgiBot.

FAQ (Frequently Asked Questions)

Q1: What is “embodied AI”?

A: Embodied AI refers to systems where perception, cognition, and action are integrated in physical agentsrobots that see, understand language or context, and perform motor tasks autonomously.

Q2: What are Vision-Language-Action (VLA) models?

A: VLA models combine vision-language understanding with real-time action control. Prominent examples include Helix, GR00T N1, and Being-0, each adopting a dual system (perception + motor) architecture.

Q3: What types of service robots were showcased at WRC 2025?

A: Robots spanned healthcare (pharmacy dispensing, prosthetics), logistics (warehouse handling, deliveries), rescue, inspection, entertainment, and even culinary services inside the Robot Mall.

Q4: What is Robot Mall?

A: The Robot Mall is the world’s first 4S store for humanoid robots serving display, sales, service, spare parts. It features dozens of robot models (Tiangong, G1, Walker S) and live demos.

Q5: How is China leading in humanoid robotics?

A: Through massive manufacturing capacity, patent output, a strong industrial ecosystem (e.g., Beijing E-Town), strategic partners like JD.com, and demonstrative events like WRC 2025, China is a global powerhouse.

Conclusion

The World Robot Conference 2025 spotlighted the era of embodied AI and humanoid robotics, where intelligent robots move beyond laboratories into healthcare, logistics, entertainment, and daily life. With foundational models like Helix, GR00T N1, and Being-0 breaking boundaries, and ecosystem innovations such as Robot Mall and strategic industry partnerships, we are witnessing a transformative leap in human–machine synergy.

China’s leadership in humanoid hardware and embodied AI systems signals a future where smart robots are not just assistants, but collaborators. For businesses, scholars, and enthusiasts, WRC 2025 served as both a showcase and a launching pad for the embodied future.