Open-source tendon-driven humanoid robotic hand developed at NYU (Generalizable Robotics and AI Lab, Prof. Lerrel Pinto). 15 DOF across 5 fingers, 11 Dynamixel servo motors in forearm. 3D-printable from PLA/TPU, matches adult human hand proportions (~18cm). Costs ~$1,300 to build; kit version available from $549 via WowRobo partnership.
Primary use case: Dexterous manipulation research; low-cost humanoid hand for academic and developer use
Price
$549
Locomotion
Stationary (robotic hand/end-effector only)
Max Lift/Arm
6 kg
Price
$549
Availability
Available as open-source DIY kit ($1,300 components) and commercial kit starting at $549 (WowRobo)
Primary Use Case
Dexterous manipulation research; low-cost humanoid hand for academic and developer use
Humanoid robots dance to K-pop hits during technology showcase in New York cbs19.tv
Researchers from York University in Toronto have developed a new stereo vision algorithm called Convergent Binocular Stereo (CBS) that enables humanoid robots to perceive depth more like humans by using cameras that converge on targets rather than remain parallel. The algorithm processes images through a five-level Gaussian pyramid, matching SIFT features and using Gabor filters to estimate both horizontal and vertical disparities for more accurate 3D depth mapping. Tested on the CBS-BM dataset of 49 scenes, CBS achieved lower mean depth errors than conventional and deep-learning stereo methods, particularly excelling in scenes with repeated patterns where it reduced error by about 0.8 meters compared to the next-best method.
Humanoid robot is heading to a New York classroom