Chinese Robot Makers' Regret: A Better Brain and More Data Would Help

Challenges in Embodied AI Development

Experts at the World Artificial Intelligence Conference (WAIC) highlighted several challenges that are hindering the progress of embodied AI, particularly in improving robots' interactions with the physical world. According to industry insiders, Chinese robotics companies face significant hurdles due to a lack of sufficient data and an effective "brain" for their products.

Wang Xiaogang, co-founder of SenseTime and chairman of its robotics spin-off Ace Robotics, emphasized that the most critical challenge is to "link hardware, data, models and real-world scenarios into a closed-loop iterative system." He noted that while a vast amount of training data is collected from human demonstrations, this data must be effectively translated into better embodied AIs. This requires joint optimization of hardware design, data-collection methods, and physical structure.

The limited deployment of robots across real-world scenarios remains another issue. Wang pointed out that the key question is how to unlock these scenarios and replicate them at scale. The availability of multi-modal data about the physical world is far from adequate compared to that used in large language models, according to Yao Maoqing, partner and senior vice-president of AgiBot. This creates one of the bottlenecks in the current training of world models expected to allow next-generation humanoid robots to model and navigate their surroundings.

Zhang Zhengyou, chief scientist of Tencent Holdings and director of the firm's Robotics X Lab, stressed the importance of integrating insights from cognitive science and neuroscience rather than relying solely on large-scale data training. Another challenge identified by experts is the gap between the video content fed to robots and the real-world data they need.

Bridging the Gap Between Brains and Bodies

Other experts pointed out a weak link between the brains and bodies of humanoid robots in physical environments. Zhang described today's smartest AI as a "brain in a tank," emphasizing that true intelligence emerges when language, vision, spatial understanding, physical control, and environmental feedback work together.

Chinese firms, ranging from tech giants to start-ups, are racing to break through bottlenecks in the fast-growing robotics sector. This sector is pushing for mass production and accelerated commercialization through wider adoption across various industries this year.

Ace Robotics introduced its action-oriented world model Kairos 3.1 during the conference, aiming to unify generative, physical, and cognitive intelligence rather than developing those capabilities separately. Wang explained that combining these capabilities in one model allows for exceptional efficiency, enabling self-reflection and self-evolution. When the model makes a mistake, it understands why it failed and knows what corrective steps to take next, creating a true closed loop.

Tencent, meanwhile, released a suite of embodied AI offerings designed to connect a robot's perception, reasoning, and movement. The line-up includes Hy-Embodied-VLM-1.0 for understanding its surroundings, Hy-Embodied-RxBrain-1.0 for reasoning and planning, and Hy-Embodied-VLA-0.5 for translating plans into physical actions.

Tencent also introduced Apexio, an always-on embodied agent, and TairosAgent, a framework for coordinating models, skills, and hardware across different robots. Zhang explained that RxBrain aims to bring together the ability to "reason" and the ability to "imagine."

The embodied AI models were part of Tencent's full-stack embodied intelligence solution, which spans from cloud infrastructure to models, platforms, and applications. Tencent has always focused on connections "between people [today], and in the future between people and robots, as well as among robots themselves," Zhang said.

Share :