Back to top
Guiding Hand · Feb - May 2022
Haptic Spatial Guidance for the Visually Impaired
Summary
I build upon a novel wrist-worn haptic glove by tweaking the original design and creating a novel localisation system, and construct research to test the capabilities of the human-machine system that has applications assistive technology for visually impaired persons.
Video
Introduction
Apple's Precision Finding Feature | Source : Apple
Motivation
Visually impaired people need a non-visual spatial guidance system.
Current consumer tech such as Apple’s AirTags utilize a technology called Ultrawide Band that can accurately locate objects in 3D space upto an accuracy of upto 10cm. However, current guidance systems only use visual means to guide users to that location. Visual means are rendered useless when visually impaired people try to use them.

The problem with audio as a means of guidance is the lack of headphones or earbuds compromises privacy. Conversely, wearing headphones hinders people’s ability to hear their surroundings, which visually impaired people heavily rely on. They also have to hold their phone up the entire time in order to receive information.
Haptics
Haptics refers to the technologies that deliver information through the sense of touch.
They are discreet and can be applied to any part of the body, making them ideal candidates for such a guidance system. Wearable haptics also free up the hands for other interactions.
Only one haptic motor in smartphones | Source : iFixit
Haptic Spatial Guidance Systems
Mobile phones just don’t make the cut.
Spatial guidance systems need to communicate two things to the user - DISTANCE and DIRECTION. Mobile phones have only one haptic motor and can be said to have a single haptic channel. A single haptic channel can only communicate one dimension - distance. In order to communicate direction and distance, we need at least 4 haptic channels - three directional channels for three axes of movement and one distance channel.
The challenge is to communicate distance and direction through haptics.
Vibrotactile feedback is applied to various body parts depending on the task and its effects have been studied - using a torso-worn vest (Ertan S. et al.), a wrist worn device, (Weber B. et al.), a head mounted haptic display (Kaul O. et al.), a waist worn haptic belt (Katzschmann R. et al.), a haptic necklace (Matsuda A. et al.), a fingertip vibrotactile glove (Elvitigala et al.), and so on. Since I was trying to achieve very precise vibrotactile guidance in 3D fordexterous tasks, it made sense to go for a wrist or palm wearable device.
Previous Work
Haptic navigation devices are usually made to work over longer distances and in 2D.
Concept Selection
Existing Solution
The TactileGlove can communicate distance and direction using haptics.
The TactileGlove is a wearable device developed by Günther et al. that has a matrix of vibration motors on the front and back of the palm. Spatial guidance using this device was tested using invisible targets within an arm's reach and showed promising results. It only lays the groundwork, however, for the research I build upon.
I developed my own prototype called the Guiding Hand
It uses the same layout of haptic motors as the TactileGlove and can guide the user to invisible targets through haptics. The way the Guiding Hand works is that there are pre-programmed targets that it can guide your hand to. It activates the vibration motors in the direction of the target. Check out how it works (headphones recommended):
Design Detailing
To make interactions feel natural and intuitive, we can design them using Metaphors from the real world.
Elastic Metaphor
The vibration gets weaker the
closer the hand gets to the target
Magnet Metaphor
The vibration gets stronger the
closer the hand gets to the target
Demonstration using directional audio. Headphones recommended.
Initial Hypothesis
Elastic might prove to be a better interaction 🤔
The flaw I see in the Magnet metaphor is that if the hand is far away from the target, then there is weaker vibration stimulus and so nudging the user in the correct direction requires a large chunk of attention. The Elastic metaphor starts out with a strong stimulus and gets weaker as the hand approaches the target. I hypothesized that this interaction will provide clear information right in the beginning, even if the hand is far from the target and ultimately increase confidence in moving the hand to the target direction.
Degrees of Freedom
Degrees of freedom refers to the number of axes that guidance will be simultaneously active in.
The scope of this project was limited to 1DOF and 2DOF only. Since the TactileGlove paper doesn’t study the effects of DOF and since it’s a detrimental detail in this interaction, the research question was chosen to be ‘How do degrees of freedom affect haptic spatial guidance?’
How much control is too much control?
It is easy to visually comprehend a lot of simultaneous data, but it’s difficult to focus on simultaneous non-visual stimuli. The more degrees of freedom a system has, the harder it is to understand.
2 Degrees of Freedom
Activates haptics simultaneously along only two axes first, and then along the remaining axis.
1 Degree of Freedom
Activates motors only along one axis regardless of where the user’s hand is along the other two axes.
Instead of user testing all possible routes, one can pick an optimal route based on logic and intuition. 3DOF or Z Guidance in 2DOF may not even be needed in a practical scenario. For example in case of a vertical guidance plane, the user can be informed of the guidance being set to vertical and then their hand can be guided in 2DOF. After they hit the target, they can simply be asked to reach forward. In case of a horizontal guidance plane like a drawer or a table, the same can be done ending with asking the user to reach straight down. Similarly, in 1DOF, the possiblity of doing Z first can be eliminated since it’s counterintuitive.
Elimination
Selecting the optimal route can be a design decision.
User Testing
The Guiding Hand was tested on 14 participants
= 40 Tests
For each test, the participants were asked to hover their hand in the middle of a testing booth and the guidance was turned on. They would follow the vibration and after they felt that the target was reached, the guidance was stopped and the deviation from the actual target as well as the time taken to reach this location was recorded. A qualitative study was also conducted after all tests.
Confirmatory Clicks were disabled on purpose
The Guiding Hand can give a haptic ‘click’ when the user reaches the intended target. While conducting research, however, this was disabled and people were asked to guess where the target was in order to study how accurately the vibration pattern alone can guide the hand.
Results
Here's what I found.
1 Degree of Freedom
Target Circle
Diameter of the circle that users may (statistically) confidently hit. The lower, the better.
1 Degree of Freedom
Time
Mean time taken by the users to reach the target. The lower, the better.
2 Degrees of Freedom
Target Circle
Diameter of the circle that users may (statistically) confidently hit. The lower, the better.
Ease and Clarity
Clarity means the distinctness of stimulus while ease means cognitive simplicity. In both aspects, the higher the better.
2 Degrees of Freedom
Time
Mean time taken by the users to reach the target. The lower, the better.
...which means...
Separating degrees of freedom leads to greatly improved accuracy at marginal or no speed tradeoffs.
1DOF 🎯 >> 2DOF
There is a huge difference between XY Separate and XY Together. The target circle diameter for 1DOF is about 36cm, whereas that for 2DOF is 77cm. That is close to double the 1DOF diameter, which means the accuracy is half that of 1DOF.
Elastic 🤝 Magnet
There is no significant difference between elastic and magnet metaphors. For Elastic, doing XY Together is much faster than doing X and Y Separately. For Magnet, the inverse is true. The fastest is doing XY together using Elastic.
User Feedback and Learnings
“The vibration is too weak”
By virtue, the vibration is supposed to get weaker as the hand approaches the target but it turned out to be near imperceptible.
Elastic Metaphor
Magnet Metaphor
“It is hard to know when the vibration is the strongest”
Since there is no reference for the strongest vibration, it can be confusing to tell when the vibration has peaked.
“There is a lot of ambiguity around the target”
As the stimulus is low around the target, it is naturally difficult to tell when the target is reached, especially due to the absence of confirmatory clicks.
“Strong vibration can be overwhelming”
Especially in 2DOF, strong vibration along either axis can muddle the vibration for the other axis, and it overall becomes confusing to comprehend.
“I have to cross check if I understand it correctly”
Due to the lack of distinct stimulus, participants mentioned intentionally going in the wrong direction just to get more stimulus about the correct direction.
“I’m only looking for the crossover point”
There is a distinct ‘crossover point’ at the target since the side of stimulus changes as the hand crosses the target. Participants had gotten used to ignoring the actual vibration pattern and looking only for this crossover.
How to improve the Elastic Metaphor
As previously mentioned, the clicks were intentionally removed to study the metaphor in isolation, but adding them back might alleviate the ambiguity and improve the speed of the Elastic metaphor.
How to improve the Magnet Metaphor
Apart from calibrating the vibration pattern to suit human perception, since we’ve concluded that XY Separate is a better guidance method than XY Together, we can dismiss the need to proceed further with XY Together.
Haptics
“I associate vibration with error”
One person said that they associated vibration with error, and so the Elastic metaphor was more intuitive since they tried to go in the path that ‘reduces error’
“I can’t tell which motor is vibrating”
Participants were occasionally having difficulty telling which direction the Guiding Hand wanted them to go in.
Mixed reviews
About 5 participants said it was easier to tell strong vibrations but not weak vibrations, whereas 4 participants said constant strong vibrations were overwhelming. Some participants said it was the easiest to tell when the vibration crossed over between up and down motors whereas Y was harder than X for a few others.
“XY is much harder than just X or Y”
Even some participants that preferred Magnet said that Elastic is easier in XY Together because the strong feedback along both axes in Magnet is confusing. XY Together was overall much harder than just X and Y.
How to improve the Haptics
Both metaphors have received feedback that the vibration is either too strong or too weak; this may be solved by mapping the vibration intensity to a nonlinear curve such that the transition from weak to strong and vice versa feels linear and consistent to the user.

We can also experiment with changing the mapping of vibration intensity from absolute to relative, such that no matter how far the target is, the guidance always starts with max vibration and ends with null vibration.

The layout can be tweaked so that the left and right motors are actually on the sides of the palm instead of the dorsum, the back motor could be on the wrist and the front motor could be near the knuckles or the middle finger for greater differentiation.
Conclusion
Wrapping Up...
The resolution of the current system is 36cm and it can be used to differentiate objects larger than that in a matrix of similarly sized objects. This study does not, however, study the factor of human intelligence that a visually impaired person would have to differentiate smaller objects by touch. There is scope for that study along with the development of a better prototype.
Full Project Report →
Credits
Guide
Prof. Anirudha Joshi
Project Details
DEP406 — B.Des Design Project 2 @ IDC School of Design, IITB
Atish Waghwase
© 2024. All rights reserved.