Study Unveils Robotic Hand That Matches Human Grip with 93 % Accuracy

Researchers have developed a robotic hand that closely mimics human grasping, achieving a 93 % success rate across diverse objects—a significant step forward in adaptive robotic manipulation.

ADAPT robotic hand (Adaptive Dexterous Anthropomorphic Programmable sTiffness)
ADAPT robotic hand (Adaptive Dexterous Anthropomorphic Programmable sTiffness). Image Credit: CREATE Lab EPFL

In a recent Nature article, researchers introduced the Adaptive Dexterous Anthropomorphic Programmable Stiffness Hand (ADAPT Hand)—a robotic hand designed to closely mimic the way humans grasp objects. The hand incorporates distributed compliance throughout its skin, fingers, and wrist, allowing it to adapt to different shapes and forces much like a human hand.

During testing, it achieved a 93 % success rate when grasping 24 diverse objects and showed a 68 % similarity to human grasp types. Thanks to its passive adaptability, the ADAPT Hand can perform robust, human-like manipulation without relying on complex control systems, completing over 800 successful grasps and approaching theoretical performance limits.

Why Existing Robotic Hands Fall Short

Human hands perform versatile manipulation by distributing compliance across skin, joints, and muscles. This structural flexibility allows for adaptive grasping with minimal cognitive or computational effort. Prior robotic hands attempted to replicate this either with soft materials or simplified mechanical structures, but often fell short, sacrificing strength, generalizability, or full-body coordination.

The ADAPT Hand was designed to close this gap. It combined a human-like structure with tunable, spatially distributed compliance, allowing researchers to systematically evaluate how biomechanically inspired design influences manipulation across varied tasks.

System Overview

The ADAPT Hand was constructed using 3D-printed PLA and TPU, combining rigid and flexible elements. It featured 20 joints across four fingers and a thumb, actuated by 12 servo motors through tendon cables.

Key design elements included:

  • Adjustable-stiffness MCP joints with replaceable springs
  • TPU flexures at the PIP and DIP joints for passive compliance
  • A compliant linkage enabling abduction–adduction motion in the MCP
  • Modular skin in soft (EcoFlex) and rigid (PLA) materials with identical geometry

Motion control relied on a custom “signal mixer box,” which used linear potentiometers to record and replay tendon displacement waypoints. The hand operated on a Franka Research 3 robotic arm, running under impedance control to facilitate smooth interaction.

Experimental Methods

The researchers tested the ADAPT Hand across three primary interaction tasks:

  1. Finger sliding measured contact forces during motion across instrumented surfaces
  2. Knob turning assessed grip and dexterity across various torques and diameters
  3. Finger gaiting evaluated coordinated finger movement for maintaining object retention

Human hand stiffness data guided design targets. The hand’s performance was benchmarked against human baselines using standardized grasp taxonomies and the Kapandji test. All in all, it successfully executed all 33 human grasp types and 10 standardized postures. Further analysis categorized performance across five grasp patterns, tested using 24 varied objects.

Results

The ADAPT Hand’s tendon-driven system demonstrated robust, passive adaptability. Soft skin outperformed rigid variants in sliding tests, generating twice the shear force and improving knob-turning performance by 52 %. The compliant MCP joints enabled natural finger movement using just three programmed waypoints, showing 2.4 times greater force consistency than rigid configurations.

In full-hand trials, the system completed 845 uninterrupted grasps over 16 hours with a 97 % success rate. It exceeded theoretical geometric tolerance limits for object displacement and adjusted grasp type based on object size, favoring fingertip grasps for small items and power grasps for larger ones. These behaviors emerged from mechanical compliance, not pre-programmed logic, highlighting the role of physical design in enabling intelligent interaction.

Despite its performance, the hand had limitations. Its passive compliance made it less effective in force-intensive tasks or scenarios requiring precise load balancing across fingers. Failures typically occurred when external forces exceeded the system’s adaptability.

The study suggested future improvements could include integrated sensor feedback and biologically inspired control strategies such as shared autonomy and coordinated sensory-motor feedback. Enhancing wrist mechanics and developing more generalizable benchmarks for manipulation complexity were also recommended.

Conclusion

The ADAPT Hand demonstrated that carefully distributed, tunable compliance can significantly improve robotic grasping performance. By mirroring the mechanical structure of human hands, the system achieved robust, versatile manipulation without the need for sophisticated control software. Its 93 % success rate across diverse objects and 68 % alignment with human grasp types made a compelling case for embedding intelligence in mechanical design.

While further development is needed for precision and force-heavy tasks, the study offered a strong foundation for physically intelligent robotic systems, where form and function work together to handle complexity.

Journal Reference

Junge, K., & Hughes, J. (2025). Spatially distributed biomimetic compliance enables robust anthropomorphic robotic manipulationCommunications Engineering4(1). DOI:10.1038/s44172-025-00407-4. https://www.nature.com/articles/s44172-025-00407-4

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Nandi, Soham. (2025, May 22). Study Unveils Robotic Hand That Matches Human Grip with 93 % Accuracy. AZoRobotics. Retrieved on May 22, 2025 from https://www.azorobotics.com/News.aspx?newsID=15971.

  • MLA

    Nandi, Soham. "Study Unveils Robotic Hand That Matches Human Grip with 93 % Accuracy". AZoRobotics. 22 May 2025. <https://www.azorobotics.com/News.aspx?newsID=15971>.

  • Chicago

    Nandi, Soham. "Study Unveils Robotic Hand That Matches Human Grip with 93 % Accuracy". AZoRobotics. https://www.azorobotics.com/News.aspx?newsID=15971. (accessed May 22, 2025).

  • Harvard

    Nandi, Soham. 2025. Study Unveils Robotic Hand That Matches Human Grip with 93 % Accuracy. AZoRobotics, viewed 22 May 2025, https://www.azorobotics.com/News.aspx?newsID=15971.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.