A soft-packaged and portable rehabilitation glove capable of closed-loop fine motor skills

Each year, there are over 7 million new patients with hand disabilities worldwide after a stroke¹. Due to impaired neural pathways², afflicted patients lose the ability to make precise, voluntary and coordinated movements of fingers—referred to as fine motor skills (FMSs). Neuroplasticity theory suggests that repetitive and consistent practices of hand motion can rewire the neural pathways that modulate hand functions^3–6. Therefore, patients usually take intensive rehabilitation training with professional therapists in the hospital^7,8. However, traditional rehabilitation therapy often involves high medical costs for patients and heavy workloads for therapists⁹.

To address this issue, recent decades have seen substantial interest in developing robotic-assisted technologies that help patients perform rehabilitation training^10–17. Consisting of rigid structures (for example, metallic rods), many existing rehabilitation devices (for example, ExoK’ab exoskeleton¹⁸ and Haptic Knob end-effector¹⁹) are still bulky and heavy, limiting rehabilitation training to only hospitals and possessing potential risks of injuring the finger²⁰. As a result, soft hand rehabilitation gloves are rapidly emerging owing to their lower weight and safer human–machine interaction^21–27. For example, hydraulic/pneumatic^28–34 and motor-cable^35,36 soft gloves that use soft tubes and cables to bend fingers have been widely studied and commercialized. Despite the reduced weight of the glove itself, external hardware (for example, pump + tubes, motor + cables) is still required for actuation^32–34, making the entire system heavier^37–42 (Supplementary Table 1).

Recently, shape-memory-alloy (SMA, for example, TiNi alloy) actuators represent a prominent candidate for creating a portable glove system. The driving force originates from the heating-induced transformation from the martensite phase to the austenite phase. Due to their high power-to-weight ratio, large deforming capability, relatively low driving voltage and noiseless actuation, SMAs have been extensively adopted as actuators in robotics fields^43–51. In particular, several groups have reported potential applications of SMAs in hand rehabilitation^52,53. However, existing rehabilitation gloves are still facing the grand challenge of accomplishing FMSs with precision^14,54–56. The main reasons are twofold⁵⁴. First, most gloves are not equipped with sensors that are directly mounted to impaired fingers to precisely capture their deformation. For example, some commercial products (for example, Vrehab-M2 and Mirror Hand) use bending sensors...