A new soft sensing system could help humanoid robots move their hands with far greater precision in delicate, human-like tasks. The study introduces a dexterous robotic hand equipped with omnidirectional bending sensors that can track both pitch and yaw at the finger joints, allowing the system to perceive complex finger posture in real time. By combining flexible sensing with a rigid-soft hand design, the researchers created a platform that not only moves more naturally but also performs demanding actions such as using scissors, operating a mouse, and playing the piano with improved control and stability.
Robotic hands have made major progress in grasping and pinching, but many still struggle with the finer motions that make the human hand so versatile. One key limitation is proprioception: while human fingers constantly sense their own position and movement, most humanoid hands remain weak at perceiving posture across multiple degrees of freedom. Existing soft sensors often detect only one bending mode or suffer from coupling problems when fingers flex and move sideways at the same time. This leaves a gap between robotic grasping and true dexterous manipulation. Based on these challenges, deeper research was needed into soft sensing systems capable of decoupling and accurately tracking multidirectional finger motion.
Researchers from Zhejiang University, Hangzhou Dianzi University, and Lishui University reported (DOI: 10.1038/s41378-026-01179-3) the work in Microsystems & Nanoengineering in 2026. The study presents a humanoid dexterous hand designed to solve a central problem in advanced robotics: how to give robot fingers a reliable sense of their own posture during complex motion. By embedding a new omnidirectional soft bending sensor into the hand, the team enabled real-time perception of both flexion and side-to-side movement in delicate manipulation tasks.
The hand features 18 active degrees of freedom and five rigid-flexible fingers, with each finger integrating a soft optical sensor built from segmented PMMA fibers, a trichromatic LED, and a chromatic detector. The design works by tracking how red, green, and blue light attenuate differently as the sensor bends. Because the fiber layout separates responses to pitch and yaw, the system can decouple the two motions instead of mixing them together. The paper reports strong repeatability over 100 cycles, with RMSE values of 2.1%, 1.9%, and 3.2% across the three optical channels. Under single bending, the average measurement error was only ±2.13° for pitch and ±2.34° for yaw. Crosstalk remained low: pure yaw contributed 3.2% to pitch, while pure pitch contributed 4.1% to yaw, with signal-to-crosstalk ratios of 50.68 dB and 30.81 dB, respectively. The team then moved beyond bench testing and demonstrated the hand in three visually compelling tasks—cutting with scissors, clicking a mouse, and playing piano keys—showing closed-loop posture control in actions that require subtle coordination rather than simple gripping.
The researchers suggest that the real advance is not just a new sensor, but a new way of giving robotic hands a more human-like internal awareness of motion. In their conclusion, they emphasize that the integrated rigid-soft design supports natural movement, while the sensing system delivers the stability, repeatability, and multi-DoF posture perception needed for complex operations. That combination could make future humanoid hands more capable in tasks where precision matters most.
This work points toward robotic hands that are not only stronger or faster, but more skillful. Better posture perception could improve humanoid robots used in service settings, industrial assembly, rehabilitation devices, and other environments where fingers must adapt to fragile or highly varied objects. The study’s demonstrations also hint at broader possibilities in human-robot interaction, where smoother and safer hand motion is essential. By showing that soft optical sensing can remain accurate while supporting complex multidirectional movement, the research moves robotic manipulation closer to the responsiveness and finesse of the human hand.
###
References
DOI
Original Source URL
https://doi.org/10.1038/s41378-026-01179-3
Funding Information
This research was supported by the National Natural Science Foundation of China (No. 52475573), the Natural Science Foundation of Zhejiang Province (No. LTGY23E050002), the National Key Research and Development Program of China (No. 2023YFC2811500), the Science and Technology Innovation Project of the General Administration of Sport of China (24KJCX074), the Key Research and Development Programme of Zhejiang (No. 2024C03259, No. 2023C03196), and the Fundamental Research Funds for the Central Universities.
About Microsystems & Nanoengineering
Microsystems & Nanoengineering is an online-only, open access international journal devoted to publishing original research results and reviews on all aspects of Micro and Nano Electro Mechanical Systems from fundamental to applied research. The journal is published by Springer Nature in partnership with the Aerospace Information Research Institute, Chinese Academy of Sciences, supported by the State Key Laboratory of Transducer Technology.
Microsystems & Nanoengineering
Not applicable
Soft sensor for omnidirectional posture perception in humanoid dexterous hands
12-Feb-2026
The authors declare that they have no competing interests.