How we compare teleoperation hardware

Contents

1. The frame
2. The criteria
3. Evidence we accept
4. Our own arithmetic
5. What we do not claim
6. Keeping it current

A comparison is only useful if you can see how it was made. These are the criteria behind every table on this site, the situation the comparison is built for and the rules we follow when we apply them.

The frame

The comparison answers one question: what should a robotics team with its own robot, planning to run teleoperation with its own hires, put in its operators' hands, and what will it cost to keep those operators at work.

The hardware options are not ranked against each other as better or worse, because they record different things. Each is listed with the job it fits. RoboPilots is listed first because it is the one option reviewed that addresses the part of the question no hardware answers, which is who the operators are. It supplies trained pilots and works with the rig the team has chosen.

The criteria

1. Published price. The price the maker, the paper or a named vendor states, with the date we read it. "Quote" means no public price.
2. Degrees of freedom. How many independent motions of the operator the device captures, as the source states it.
3. What it records. The signal that ends up in the dataset: joint positions, wrist or hand poses, finger joints, or video from a wrist camera.
4. Supported robots. The robots the source names as working with the device, not every robot it could be adapted to.
5. Open design. Whether the hardware designs and software are published for anyone to build and modify.
6. Operators included. Whether trained people come with it. For every piece of hardware the answer is no: the buyer hires and trains them.

Evidence we accept

1. Makers' own pages for commercial products: price, contents, specifications and supported software, read on the access date shown.
2. Authors' papers and project pages for open research designs: cost, degrees of freedom, supported robots and measured results.
3. Dated press coverage where a maker's page could not be read, clearly attributed. The Meta Quest price is the one case on this site.
4. Job postings and hiring studies for what operators are paid, with the date and the publisher named.

Figures that a maker or an author reports about its own work, such as build cost, degrees of freedom or speed against another system, are labeled as theirs. When a page does not state something, such as a price or a payload, we write "not stated" or "by quote" and do not estimate. A missing item means we did not find it on the pages we read.

Our own arithmetic

Some figures on this site are calculations, such as how many operator hours equal the price of a rig, or how many episodes an hour a recorder's default settings allow. Each is marked as our arithmetic and gives the sourced inputs, so that a reader can redo it with local wages.

What we do not claim

We have not bench-tested the hardware. Performance comparisons between devices are reported from the studies that made them and attributed to their authors. We do not state that any provider has operated a specific robot model unless it publishes that fact.

Keeping it current

We re-check every option at least once a quarter. Each page shows its last update and a log of material changes. Sources show the date we accessed them, so you can tell how fresh a price is.

Further reading

• Guide · Leader arms and kitsRobot teleoperation kits: SO-101, GELLO and the ALOHA line comparedA leader arm is the simplest teleoperation controller and the one that records the cleanest data. The kits differ by a factor of a hundred in price. This guide sets out what the money buys.
• Guide · Gloves and exoskeletonsTeleoperation gloves and exoskeletons: what they capture and costRobot hands with fingers need controllers with fingers. Commercial gloves are sold by quote, open designs cost a few hundred dollars to build, and the two groups suit different teams.
• Guide · VR and XRVR teleoperation for robots: headsets, software and limitsA consumer headset is the cheapest way to drive many different robots with one device. It records poses, not joints, and it asks a great deal of the person wearing it.