Humanoid Guide introduces Embodied AI’s “AIR”

Humanoid Guide published an article introducing Embodied AI’s “AIR” on November 6, 2026.

The publication of an article introducing AIR does not in itself verify the robot’s performance or practical utility. In evaluating physical AI, descriptions of a product’s appearance and functions should be considered separately from results obtained in actual tasks. See the original source for details on AIR’s specifications and evaluations. Our related articles cover robot-learning platforms and evaluations of real-world tasks separately. [Hugging Face releases robot-learning platform LeRobot v0.7.0, supporting conversational control and large-scale data training](/en/news/hugging-face-releases-lerobot-v0-7-0-robotics-training-platform-adding-4e8d3f) introduced a platform that uses about 300TB of open Physical AI data on the Hub. Meanwhile, Panasonic HD and others jointly develop robot manipulation AI “T-Rex,” with a 65% average success rate across 12 tasks reported an average success rate across 12 tasks involving tactile sensing. When reading the AIR article, it is also important not to conflate descriptions of a learning platform with reproducible results on individual tasks. As Asumekku begins handling 8.3MP GMSL2 camera modules shows, the specifications of input components such as cameras are also a separate point to verify when considering robot perception. However, the articles cited here cover different subjects, and their figures cannot be directly compared to evaluate AIR’s capabilities. Going forward, it will be important to see whether information on AIR specifies the tasks, evaluation conditions, number of trials and responses to failures. There is no basis for interpreting the publication of an introductory article as evidence of mass production or field deployment.