Hyppää sisältöön
TACCU

Tampere Accessibility Unit

Accessible Design 3D Printing and Alternative Inputs for Devices

Some common assistive products do not usually come from a catalogue. Some of the most useful things in a classroom are customized, modified or made locally: a grip shaped to one student’s hand, a stand that holds a tablet at the right angle, a switch a student can reach.

This page covers designing and making assistive products, including with 3D printing, and the alternative input devices people use when a standard keyboard and mouse do not work for them.

Three things that usually get confused

Universal design changes the environment or product so it works for the widest range of people from the start. Reasonable accommodation is an individual adjustment inside an environment that already exists. Assistive technology is a device an individual uses to get past a barrier the environment still presents.

Universal design reduces how often accommodations and assistive technology are needed. It does not remove the need.

A ramp does not remove the need for a wheelchair, and a well-designed course does not remove the need for a screen reader.

Universal design rests on seven principles: equitable use, flexibility in use, simple and intuitive use, perceptible information, tolerance for error, low physical effort, and enough size and space to approach and use. Design for All, the European version of the same idea, adds one more: involve end users at every stage in the design process.

Why designing without disabled people fails

There is direct evidence for this from the maker community itself. A study presented at CHI 2015 catalogued 363 assistive technology designs shared by 273 designers on Thingiverse, the largest open repository of printable files. More than 76% of those designers reported no disability. Fewer than 1% had any training in assistive technology. Only 13 had made the device for their own use.

That matters because assistive technology designed for people rather than with them tends to be abandoned. The classic study of abandonment, by Phillips and Zhao in 1993, found four predictors: the user’s opinion was not taken into account; the device was obtained without proper assessment; it performed poorly; and the person’s needs changed. Three of the four are process failures, not technical ones.

There are ways to avoid this and most of the frameworks listed on the Procurement and acceptance page share a simple rule that requires assessment of the person and the context first, before choosing the technology.

Making assistive products locally

3D printing lets a university make a custom aid in hours instead of importing one in months. It works well for grips and handles, stands and mounts, keyguards that stop accidental key presses, positioning aids, replacement parts, and tactile teaching models.

Tactile models are worth a mention of their own. A study comparing blind students given 3D printed models with students given braille-labelled tactile graphics found that students with models handled them from different angles and rearranged the pieces, while students with flat graphics tended to recite the labels. For a molecule, a bone or a map, a printed model teaches something a raised diagram cannot.

Safety is one thing that needs to be taken into consideration and what to do depends on the case. A systematic review of 18 studies on 3D printed prostheses in developing countries found real advantages in cost and fit, alongside a World Health Organization figure that only 5 to 15% of people in lower-income countries have access to a prosthesis at all. Its conclusion was a caution: the evidence is fragmented, and much more research is needed before drawing conclusions about safety.

Before printing anything, agree five things: who assessed the person and how; who is accountable if the device fails; whether the person will test and adjust it; who prints the replacement, and from what supply of filament; and whether anyone else has tried the design.

Designs can be found in many places including the Assistive Technology groups and the Makers Making Change collection on Thingiverse, and in the NIH 3D Print Exchange. Treat every downloaded file as a starting point.

Alternative ways to control a computer

Start with what the device already does. Windows includes sticky keys, filter keys, mouse keys, an on-screen keyboard a switch can operate, and speech recognition. iPhone and iPad include eye tracking using the front camera with no extra hardware, plus dwell control, switch control and voice control. Android includes Voice Access and Switch Access. ChromeOS includes sticky keys, zoom and voice typing. Eye gaze control used to mean a device costing thousands of euros. On a recent iPad it is a just a setting.

Then the specialist options: adapted or large-key keyboards and keyguards; trackballs and joysticks; switches and switch scanning, where a single reliable movement operates a switch and software scans through options on screen; head pointers and head tracking; eye gaze systems; sip-and-puff controls; and voice input.

Eye gaze needs planning, not just budget. It requires calibration for each user in each session, consistent lighting, since it performs poorly in strong sunlight, a stable mount at a fixed distance, and recalibration when seating changes.

Voice input must be tested and some tools may need training. A study in PNAS tested five commercial speech recognition systems and found an average word error rate of 35% for Black speakers against 19% for white speakers. A 2025 review of speech recognition for 111 African languages found that Kinyarwanda has a large speech corpus that is not yet usable without further work, and that 63.5% of African-language datasets carry a high risk of bias from small size. Test voice input with the student, on their own device, before recommending it.

New ways to customise technology keep appearing. Large language models can already
describe images, summarise readings, transcribe speech and rewrite text at a
different reading level, and they can be adapted for specific tasks and specific
users. Brain-computer interfaces, smart wearables and eye tracking built into
ordinary consumer devices are moving in the same direction: more customisation,
lower cost, less specialist equipment.

If you want to read further, these are good starting points:

– Artificial Intelligence and Accessibility Research Symposium (W3C Web Accessibility Initiative), which brought together researchers, industry and disabled people to look at where AI helps accessibility and where it causes harm

– WIPO Technology Trends 2021, Assistive Technology, which maps where innovation in assistive products is actually happening

– The WHO and UNICEF Global Report on Assistive Technology, for the gap between what exists and what people can get

Making it last

A locally made device is only as good as the arrangement to remake it. A study of repair in low-resource settings found 28% of respondents said specialist repair was unavailable and another 47% said services were limited. In Sierra Leone, 50% of devices needed repair and 45% of users could not afford it.

Maintenance, support and lifecycle covers this in full.

The measure of a design programme is not how many devices it produced, but how many are still in use a year later.

Checklist and practices

Before making anything:

  • Keep universal design, accommodation and assistive technology distinct in your planning
  • Assess the person and context first, then choose the technology
  • Involve the person in setting requirements, testing and adjusting
  • Check whether an existing product already solves it

When making a device:

  • Agree who is accountable if it fails
  • Keep printed parts away from load-bearing, skin-contact and medical uses
  • Plan filament supply, printer maintenance and power before the first print
  • Record the design and settings so the part can be remade

When choosing an input method:

  • Check the free built-in features of the device the person already owns
  • Test voice input with the individual speaker
  • Plan calibration, lighting and mounting before choosing eye gaze
  • Use a multi-professional assessment, not a supplier demonstration

Remember:

  • Universal design reduces the need for assistive technology; it does not remove it
  • Over 76% of assistive designs shared online were made by people without disabilities
  • Three of the four predictors of abandonment are process failures
  • Free built-in features are the first option, not the last resort
  • Speech recognition accuracy varies by accent and language
  • Design for the second device, not the first

Links and references