The insufficient proprioceptive ability of bionic artificial muscles will greatly restrict the robot's ability to adaptively and precisely control in different scenarios. Inspired by the “sliding filament theory” of biological muscles, we developed an electromagnetically driven artificial muscle that integrates actuation and sensing based on a built-in liquid‒metal dual-coil system, and achieved a structural fusion of actuation and sensing functions. The artificial muscle exhibited a contraction ratio of 42%, a response time of 100 ms, and proprioceptive deformation sensing precision at the submillimeter level (0.5 mm), with no visible decay in the sensing performance after 1000 loading cycles. Through the process of touching and sliding, the finger module driven by artificial muscle could identify the different shapes and heights of objects. Furthermore, these artificial muscles were successfully used in a multi-DOF dexterous hand and could perform tasks such as gesture simulation. This research overcomes traditional limitations and features a high level of integration, offering a solution for the miniaturization and enhanced autonomy of soft robots.
Chen et al. (2026) studied this question.