Hacker News Reader

Comment by grondilu | Hacker News Reader

Why do robots rely so much on gears? There is no gear in the animal kingdom, and yet their limbs are vastly more energy efficient and agile, aren't they?

View on HN

Muscles/limbs are only 'vastly' more efficient if you consider they have large numbers of nano scale support systems constantly rebuilding them. Since we don't have nanobots, gears will be better for machines. Also, nature didn't naturally develop a axle.

Robots have bearings and simple, efficient rotary electric motors. A machinists lathe is a fantastic piece of technology. We're really good at building things that rotate.

Nature uses cellular-scale nanotechnology to build millions of microscopic linear motors. We can build linear electric motors by essentially unrolling the stator, but they're on a much larger scale.

We get better results with servomotors, either installed with some gear reduction at a pivot joint, or hooked to a ballscrew linear actuator.

Muscles/limbs also don't power-down for long periods. They're in need of constant maintenance (energy, food; waste removed) or they rot.

Is that an inherent property of muscle-like mechanisms, or just several billion years' worth of environmental bacteria evolved to consume living tissue if not constantly opposed by a immune system?

That's an inherent property of all living material. All tisssue is on the edge of thermodynamic breakdown i.e. Unlike manufactured material, resting state for biology is death).

There are functioning gears in the animal kingdom: juvenile Issus insects have their rear hind leg movement synchronized with gearing.

http://www.cam.ac.uk/research/news/functioning-mechanical-ge...

Electric motors can't generate the torque to drive something like a knee joint, without being very large and heavy.

Motor torque is limited by the current you can run through copper wire, times the magnetic field you can generate with permanent magnets. The best motors are only about 2 Nm / kg (continuous torque / motor weight). A knee joint requires torque around (weight of robot/2 * length of tibia) in order to squat, say 100 nM for a human-sized robot. Direct-drive motors would weigh 50 kg each! So in practice, you use a 20:1 gearbox and 2.5 kg motors.

Superconductors could totally change that some day.

Alas, there is no muscle in the robot kingdom! A lot of interesting work has been done recently on direct drive or minimally geared legged robots, however - check out Ghost Robotics for an example. [https://www.ghostrobotics.io/]

There is atleast one animal that uses gears as part of their locomotion: http://www.npr.org/2013/09/13/219739500/living-gears-help-th...

There are no muscles on mars either though. NASA has a different set of constraints than nature (one of them being human lifespans of time to make a workable solution).

Is there a biological analog of gear ratios? A nice thing about mechanical gearing is the ability to trade off speed and torque on the fly.

The length of the lever. For speed, long legs. For torque, short jaws.

Agility varies quite a lot. For example, a pick-and-place robot's arm has more agility than a grasshopper's hind leg, as that leg is pretty limited in its movement.

Reply on news.ycombinator.com