This open source humanoid hand is 3D-printable for $650
Most tendon-based robot hands face the same maintenance problem: the tendons that drive the fingers stretch, slip off their routing paths, and need periodic re-tensioning. A team at NJIT thinks it has a better architecture.
The team, led by PhD candidate Neel Adwani under Professor Cong Wang in NJIT's Electrical and Computer Engineering department, published a paper accepted at IEEE-RAS Humanoids 2026 describing the PHD-Hand: a Printable Humanoid Dexterous Hand that replaces tendons with rigid steel linkages.
"Unlike other linkage-driven designs of similar dexterity, our design does not require many miniature moving parts," the team says. "As a result, the hand can be fabricated easily with a budget 3D printer."
Total build cost: under $650. License: open hardware.
This is huge for hobbyists but also those trying to build low-cost robots.
What PHD stands for
PHD is Printable Humanoid Dexterous.
The design clears 3 barriers at once: cost (most research-grade dexterous hands run well above $5,000), maintainability (tendons), and accessibility (closed IP). The team's stated goal is to lower the barrier to entry for anyone who wants to build and study a dexterous humanoid hand.
How the linkage drive works
Each of the hand's 5 fingers is driven by 3 servos, all mounted off the digit itself. Instead of running a cable through the finger that must maintain tension, the PHD-Hand uses 0.6mm stainless steel spring wire running through a TPU flexure hinge at each joint. The hinge bends in 2 axes while resisting twist. Forward kinematics error stays within 1.8 degrees across the full range of motion.
The result is true bidirectional actuation: every joint can be actively opened or closed by the drivetrain alone, without return springs. Adwani described the contrast plainly: "Tendons are what make most affordable robot hands hard to keep running: they stretch, slip and need constant re-tensioning."
A tendon hand can't actively extend a finger after a grip. The PHD-Hand can. That matters for tasks like precision screwdriver turning, where you need to release and reset your grip mid-rotation without the finger passively flopping back.
The Humanoid Technology Expo is where the builders, integrators, and enterprise buyers shaping this space come together. Early registration is open.
The numbers
The hand has 15 total degrees of actuation across 5 fingers, 3 per digit. It measures around 190-200mm in length, within human hand size range and notably smaller than competitor designs that run 218-255mm. The weight is just 570 grams, and the part count per digit is 18, or 13 if phalanges and joints are printed as one piece.
Most competitor designs typically require 27 or more parts per digit.
On a single prototype logged across approximately 200 hours, only 2 drive rod failures occurred, one attributed to a defect in raw material stock. No other mechanical degradation was recorded.
That's impressive at this cost level.
What it can do
The team tested against the Feix GRASP taxonomy, which covers all 33 documented human grasp types.
Amazingly, the PHD-Hand demonstrated all 33, with noted limitations in 3 (Stick, Light tool, and Tripod grasps) due to limited flexion range and the absence of a thenar web space (the fold of skin between your thumb and hand, which turns out to be pretty important when gripping objects).
Object manipulation tasks included:
- jar lid removal
- precision screwdriver turning
- crescent wrench adjustment
- and cube rotation in-hand
The team did not conduct force measurements in this iteration ... that's flagged for a follow-up study.
Open and printable
The design is released under the CERN-OHL-P v2 permissive open hardware license. Software is MIT. Files will include CAD in STEP and Fusion 360 formats, STL and 3MF print files, and a full bill of materials.
A browser-based GUI handles jogging, calibration, preset postures, and motion recording and playback with no installation required.
The project site is at phd-hand.github.io; the GitHub repository was listed as coming soon at time of publication.
Why this matters
Dexterous hands are one of the main unsolved bottlenecks in humanoid robotics. Most commercial humanoids ship with hands that grip and release but can't manipulate objects with precision. Research-grade hands that can do more typically cost as much as a used car and require significant engineering overhead to maintain.
The PHD-Hand doesn't solve all of this, but it certainly raises the bar.
A team with a 3D printer and $650 in parts can now build a hand with 15 degrees of actuation, true bidirectional drive, and demonstrated performance across the full Feix taxonomy. In the teams' words, the design "significantly lowers the barrier to entry for using and developing dexterous humanoid robot hand functions."
I'm impressed.
Sources
In this story
Follow to get an alert when they're in the news
Humanoid (almost) Daily
The humanoid robot news that matters, with context from HumanoidMegaHub's tracking. Free, almost every day.
Keep reading
Written by
I tell stories with data and build startups. I've raised angel and VC funds, led software companies and startups, and chronicled the rise of the mobile economy. Currently I write for Forbes, host several podcasts, and am starting up HumanoidMegaHub. I was TUNE's mobile economist and Singular's VP of Insights, built the research team at VentureBeat, and managed teams creating software for partners like Intel and Disney. In past lives I built websites, e-learning platforms, and mobile apps, and, a long time ago, was named to Folio's top 100 of the media industry's "most innovative entrepreneurs and market shaker-uppers.”