Video: Dexterous robot hand learns the delicate art of silk embroidery
Video: Dexterous robot hand learns the delicate art of silk embroidery
Agilink's OmniHand 3 Ultra, trained by a master of Suzhou embroidery, has learned to split silk strands, thread a needle, and stitch with precision. The hand uses vision-based sensors and direct-drive mechanics for fine control. The article contrasts this with other robotic hands, such as LinkerBot and Boston Dynamics' Atlas hand, to explore different approaches to dexterous manipulation in robotics.
Source: New Atlas — Robotics · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
weight
630
View original evidence
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4weight
3
View original evidence
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4weight
≤8
View original evidence
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4time
≈0.3 seconds
View original evidence
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4precision
≈0.2
View original evidence
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4force
0.08
View original evidence
ct surface for tiny changes, a bit like a camera peering at the inside of a soft pad. According to the company, the fingertips measure force in three dimensions and detect deformations of 0.08 mm (0.003 in). That helps the hand tell a slipping object from one it's merely brushing. The palm adds a 300-point contact sensor. Silk is a nightmare for a robot to handle Agilink - YouTube The embroidery stunt points at a rea
Open source S5contact points
300
View original evidence
ct surface for tiny changes, a bit like a camera peering at the inside of a soft pad. According to the company, the fingertips measure force in three dimensions and detect deformations of 0.08 mm (0.003 in). That helps the hand tell a slipping object from one it's merely brushing. The palm adds a 300-point contact sensor. Silk is a nightmare for a robot to handle Agilink - YouTube The embroidery stunt points at a rea
Open source S5weight
1.4
View original evidence
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4
- paper: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
n is roughly the size of a human hand and weighs 630 g (1.4 lb). It uses a direct-drive design, meaning no tendons or intermediate transmissions sit between motor and joint. It can hold 3 kg (6.6 lb) stably, lift up to 8 kg (17.6 lb), open and close in about 0.3 seconds, and repeat positions to within about 0.2 mm (0.008 in). The robohand’s touch relies on vision-based sensors in each fingertip, which watch the conta
Open source S4
ct surface for tiny changes, a bit like a camera peering at the inside of a soft pad. According to the company, the fingertips measure force in three dimensions and detect deformations of 0.08 mm (0.003 in). That helps the hand tell a slipping object from one it's merely brushing. The palm adds a 300-point contact sensor. Silk is a nightmare for a robot to handle Agilink - YouTube The embroidery stunt points at a rea
Open source S5
Source：New Atlas — Robotics · newatlas.com