Robot Control Tech That Lets Machines Learn Their Own Bodies Enters Proof-of-Concept Stage

According to AI TIMES, Robai began a proof-of-concept test on October 19 with China's Linkerbot for AI control technology that enables robots to learn their own body structures and movements. The initial verification will apply Robai's self-exploration platform to three of Linkerbot's robot hand and arm products, with a focus on determining whether it can reduce the burden of control development and calibration even when the hardware differs. The key question in this collaboration is not whether AI can create task plans, but whether machines can independently learn control methods suited to the joint and actuation characteristics of real robots. Until now, control software often had to be reworked whenever the manufacturer or product changed, and the two companies plan to test on commercial equipment whether that burden can be eased. During the verification, the two companies will examine both whether self-learning actually works and whether initial setup and calibration burdens are reduced compared with conventional engineering methods. After the verification, they plan to review the possibility of developing an integrated solution and pursuing joint commercialization based on the results.