Underactuated Hand

• Also called: underactuated gripper, adaptive gripper, self-adaptive hand, underactuated robotic hand
• Category: Mechanical structure › Arms & end-effectors
• Tags: grippers, end-effectors, grasping, mechanism-design, tendon-driven, manipulation

A robotic hand or gripper with fewer actuators than joints that uses tendons, linkages or springs so its fingers passively wrap around and conform to the shape of a grasped object.

Definition

Underactuated hands are robotic hands and grippers that have fewer actuators than degrees of freedom. Instead of driving each joint with its own motor, a single actuator drives several joints through a mechanical transmission, such as a tendon running through the finger, a linkage or a differential, while springs or compliant joints set the default posture. The remaining degrees of freedom are not controlled directly; they are determined by contact between the fingers and the object.

The benefit is adaptive, shape-conforming grasping. When a finger closes, its base segment moves until it touches the object; once that segment is blocked, the transmission keeps driving the outer segments, which curl around the object. The hand therefore wraps around irregular shapes without needing a model of the object or precise sensing. This makes underactuated hands simpler, lighter, cheaper and more robust than fully actuated hands with a similar number of joints.

The trade-off is reduced dexterity. Because joints cannot be commanded independently, underactuated hands are good at power and enveloping grasps but limited in in-hand manipulation, where fingers must reposition an object in a controlled way. Their behavior also depends on friction, spring stiffness and contact locations, which can make precise fingertip grasps less predictable.

Underactuation is widely used in commercial adaptive grippers and in open-source designs such as the Yale OpenHand project, which publishes 3D-printable tendon-driven hands. The design principles are covered in depth in the book Underactuated Robotic Hands by Birglen, Laliberté and Gosselin.

Why it matters

Grasping objects of unknown or varied shape is one of the hardest problems in robotics, and underactuated hands solve much of it mechanically instead of through sensing and control. They make reliable grasping possible with one motor and simple control, which is why they are common on mobile manipulators, in logistics and in low-cost and open-source robots.

Key applications

• Adaptive grippers for logistics and bin picking
• Prosthetic hands
• Open-source and low-cost robot hands
• Grasping in unstructured settings such as homes and farms
• Robot hands for learning-based manipulation research

In practice

A team choosing a hand for a pick-and-place or household robot often starts with an underactuated design because it needs only one or two motors and a simple open/close command. In open-source builds, fingers are commonly 3D printed with tendon channels and flexure or pin joints, and a single servo winds the tendon. Engineers tune spring or flexure stiffness so the finger segments close in the intended order, and they test grasps on a set of representative objects to check that the hand both wraps large objects and can still pinch small ones. When learning manipulation policies with such hands, they account for the fact that the actual finger joint angles differ from the commanded actuator position, sometimes adding joint encoders to measure them.

Common misconceptions

• Underactuated means weak or underpowered — the term means fewer actuators than joints, not low grip force
• An underactuated hand cannot be controlled — its posture is set by the actuator and the contacts together, and it behaves predictably for the grasps it was designed for
• Underactuated hands can do anything a fully actuated hand can — they give up independent finger control, and with it much in-hand dexterity, in exchange for simplicity and adaptability

Who should know this

• Gripper and end-effector designers
• Manipulation and grasping researchers
• Prosthetics engineers
• Makers building low-cost robot hands

Related technologies

• Tendon-Driven Actuation
• Yale OpenHand
• Compliant Mechanisms
• Adaptive Gripper
• Soft Gripper
• Flexure Joints

Related terms

• Gripper
• Dexterous Manipulation
• End Effector

Related projects

• LeRobot: HuggingFace ML library for real-world robotics in PyTorch
• Isaac Lab: NVIDIA robot learning framework: one task API across Newton, PhysX and Isaac Sim
• NVIDIA Isaac GR00T: NVIDIA open foundation model for humanoid robot reasoning and skills
• OpenVLA: 7B-parameter Vision-Language-Action model for robotic manipulation
• Drake: MIT robotics toolbox for model-based dynamics, optimization, and control system design
• OpenArm: 7DOF humanoid arm for physical AI research in contact-rich environments