My three partners and I have be developing and selling multi-camera arrays specifically for eye tracking as well as measuring other physiological features for several years now. Our main customers are a couple university research groups, a human factors group in Lockheed, and just recently the US Air Force. In fact we just returned from a trip to Wright-Patterson installing an array in a hypobaric chamber to perform gaze tracker and pupil response for pilots under hypoxic conditions. Phase two will be a custom gaze tracker for their centrifuge. Our main features are accurate eye and face tracking up to a meter from the array, minimal calibration per subject (about 10 seconds staring at a dot), pupil response for measuring fatigue and other things, plus we can adapt the array for the client ranging from a cockpit to a large flat screen TV. We've looked into medical usage such as ALS, but we're bootstrapped based in Iowa and found the military niche as a more direct way to generate cash flow. It's ashame we can't apply this work towards people with medical needs, but we don't have the funds nor the clients to make such a pivot.
Have you thought about settinng up a subsidiary that licenses your base tech for a reasonable royalty fee and raise capital for the subsidiary to develop a medical product from it?
The risk and part of the returns there are for the investors. While it will generate additional revenue (and diversification) for your bootstrapped company allowing you to keep building and mitigate some of the risk of having a narrow (military) client base.
And if it becomes a major success (sounds like pg thinks that's possible) you'll co-own it.
One of our partners is a renowned neuro-ophthalmologist at our local university, and he's always suggested medical applications for our array. However, the combination of low volume, slim margins, and regulatory challenges (acquiring patient training sets and navigating FDA approval) makes this direction potentially perilous for us. What helps us is our human factors partner has decades long connections with NASA, FAA, and DoD which gets us a foot in the door, and allows us adopt a low volume/high margin strategy. This approach is letting us refine our technology and then we can focus on standardizing it, reducing costs, and perhaps refocus on medical applications.
I do think your suggestion is a good approach, but what's key is finding smart investment that has experience in developing medical devices.
Luxury car prices per array due to development, customization and the type of clientele. We have a mech engineer onsite to adapt the camera arrays to the environment, such as a cockpit or centrifuge, while taking into account the field of views and mounts. Yes it can handle multiple screens and curves like domes, the fun part is working out the geometry to convert gaze vectors into pixel coordinates. Pixel accuracy is a function of the number of camera (6 to 12), distance to the array, and geometry of the subject and array but we can typically discern which button / control / digit a pilot observed.
Of course cost can be substantially decreased if everything was standardized and the mech engineering was done once. BOMs (excluding the high-end desktop computer) is about $2000-$3000. My dream would be to reduce that cost by moving the computer vision to compute modules per pairs of cameras reducing the BOM to < $1000 and avoiding the desktop.
I responded to the thread but Senseye has been working on this for a while now. Originally they were working with the US Air Force to help with improving pilot training - fatigue etc.. inference with retinal reading
They have generally struggled to find funding for their eye tracking focused work, and have recently had to pivot away from the really exciting but hard to fund stuff into PTSD screening (which is important too).
I can connect you with the founder if desired via the email in my bio
I do hardware. I do software. I do computer vision. I built some software that ran on a cellphone used by LEO (law enforcement officers) to determine if the person they are quizzing is inebriated or impaired through controlled substances by examining the person's eyes and having them focus on images displayed on the phone screen. I've done eye tracking using fully custom solutions and also through a few of the off-the-shelf SDKs such as GazeSense from eyeware and a few other SDKs.
The problem is not the eye-tracking, it is reasonably easy to build robust systems that can do that easily enough, even with custom hardware under all sorts of lighting conditions. The hard part is the UX if you are trying to build something that isn't hampered by current UI paradigms.
Rapid typing and menus of custom actions with just eye movement, though fatiguing, shouldn't be hard to solve, and then render the output however you want; text, text to speech, commands issued to an machine, etc. Making a usable user interface to do anything else, that's where the rubber hits the road.
@pg, which software is your friend using? If it is anything like I've looked in to in the past, it's over-priced accessibility crap with a UI straight out of the 1990s.
Yes, UX is the key. The iPhone succeeded because they didn't just take macOS's mouse/keyboard UI and slap it under a touchscreen. They took the limitations and strengths of capacitive touch and designed a bespoke UX with new concepts everywhere.
Input modalities define platforms. Eye tracking is a new input modality and will define a new platform. It needs a whole new UX designed around its limitations and strengths. It needs a keyboard, it needs a browser, it needs copy and paste, it needs an app switcher, it needs a whole vocabulary of standard interactions and best practices. Apple has a good start in Vision Pro but they're not going to be the only ones doing UX for eye tracking. There's definitely room for other players with fresh ideas.
Another good example is the early "handwriting recognition" of Palm Pilot. They didn't try to increase the accuracy of recognizing all letters, they tried to hit the sweet spot between what they coudl reliably interpret and what users would learn/put up with.
I've got ALS (MND). Completely agree UX is the problem, gazing at a keyboard on a screen designed to stop multiple keys clogging (QWERTY) feels wrong.
Some ideas
- gesture based eye movements, maybe two sweeps on a nine by nine grid, which map directly to phonemes
- enormous 4k 75inch tv with thousands of words or ideograms or phrases
- "writing" with your eyes then doing line to text AI to clean up
- standardish keyboard with massive LLM prediction and clean UX for autocomplete/throwaway with branching options
Ideas are cheap so no clue if these work. Also Tobi split between cheap good non-hackable gaming eyetracking and medical products doesn't help. Finally, with ALS you want to communicate about different things and are more tired.
I worked on the Eye Control Keyboard that has shipped as part of Windows since version 10. You are right that having a QWERTY keyboard is wrong, in many ways.
The Windows keyboard does actually implement something similar to your first and third suggestion. You can spell words by fixating on the first letter of a word, glancing between the following letters and finally fixating on the last letter. Some people can do this to successfully and achieve good input speeds, however, it is a skill that takes some mastery.
For me the real problem comes from three places.
Firstly, having to spell out words either with some kind of keyboard or even with a Dasher-like system means word length matters, long words are harder to enter than short words. The amount of effort needed to express an idea should be proportional to how unusual that idea is, not how many letters are needed to express it; "Hello Dave, nice to see you, how are you today?" should be easier to write than "Eel shoes".
Secondly, in order to achieve some level of throughput, you need to accept that you're going to be living near an edge where typos are inevitable. On current systems, the mechanisms for making corrections are extremely disruptive to throughput, mostly involving repeatedly pressing a key to delete the last character, word or sentence.
Thirdly, similar to the second issue, revising finished text is also a fraught problem that is inadequately addressed and often requires repeated pressing of arrow keys and the like.
I am working on a solution that I believe addresses all of these issues. A solution that allows text to be created quickly using input methods slower than typing, be it eye tracking or switch access scanning. Similarly, a system that allows the same input methods to be used to review and revise text efficiently.
I will be looking to see if I can take advantage of Paul's interest in this area to help his friend and others.
Thank you as well for the encouragement. I'm currently head down working on getting my code into something that could ship as a minimum-viable-product on a variety of platforms.
> I built some software that ran on a cellphone used by LEO (law enforcement officers) to determine if the person they are quizzing is inebriated or impaired through controlled substances by examining the person's eyes and having them focus on images displayed on the phone screen.
Is this used in the field, and what level of testing and validation was required?
This is coming from a skeptical POV, but I’m genuinely curious to hear about the process. Historically, of course, law enforcement technology has a history of being anything but evidence based. It’s good that there’s finally progress away from pseudoscience like fiber analysis and smoke patterns.
But from your experience is there anything to stop LEOs from adopting new tools and technologies without adequate evaluation and standards, aside from the courts, which have such a poor track record in this space?
EEG recording is an alternative that would outlast the potential disease-related degradation of eye movements. Manny Donchin gave a brown bag at UIUC about the possibilities of using this approach to support communication by ALS patients many years ago. It's clever: they use the P300 marker to index attention/intention. I do not recall whether he and his colleagues ever commercialized the tech. I believe that this publication is representative: https://doi.org/10.1016/j.clinph.2005.06.027
I did a PhD in brain-computer interfaces, including EEG and implanted electrodes. BCI research to a big extent focuses on helping paralyzed individuals regain communication.
Unfortunately, EEG (including P300) doesn’t provide sufficient signal-to-noise ratio to support good communication speeds outside of the lab with Faraday cages and days/weeks of de-noising including removing eye-movement artifacts in the recordings. This is a physical limit due to attenuation of brain’s electrical fields outside of the skull, which is hard to overcome. For example, all commercial “mind-reading” toys are actually working based off head and eye muscle signals.
Implanted electrodes provide better signal but are many iterations away from becoming viable commercially. Signal degrades over months as the brain builds scar tissue around electrodes and the brain surgery is obviously pretty dangerous. Iteration cycles are very slow because of the need for government approval for testing in humans (for a good reason).
If I wanted to help a paralyzed friend, who could only move his/her eyes, I would definitely focus on the eye-tracking tech. It hands-down beat all BCIs I’ve heard of.
++ with mere MA in cog sci and psychology. If you really wanted to get EEG to work for typing, maybe you could train someone to map thinking about specific kinds of things to the keyboard, but that would be an extremely weird experience. Eyeball is going to have the best signal about eyeball-related motor cortex we can access.
This! Eye tracking is slow and not good - but does that just mean we need to “faster horse” it or is there another option for bridging the communication gap for people with ALS and similar diseases? I have to believe there are better answers with other tech - likely EEG (+ AI).
My family is one of the unlucky ones that has genes for ALS so I’ve watched enough family members struggle. (I’m lucky, selfishly, because I dodged the gene but I still care deeply about this).
I had this thought, but then I thought about if my friend was struggling with a problem had had practical but imperfect solutions, would I better serve them by funding highly feasible solutions that they're already familiar with, or experimental moonshots that are more likely to fail, will take longer to implement, and my friend may not even care for at all...
I worked on eye tracking hardware for Microsoft HoloLens. Several AR headsets offer decent eye tracking, including Hololens 2 and Magic Leap's ML2. I think Tobii's eye tracking glasses are probably better as a stand-alone solution though: https://www.tobii.com/products/eye-trackers/wearables/tobii-...
Agreed, the eye tracking itself is really a mostly-solved problem (Tobii are indeed leaders in the area). It's how it's used that matters - and as mentioned above, it's likely that it's the usability/interface that needs work.
PS VR2 uses Tobii's tech to do eye tracking, it is mostly being used for foveated rendering but some games also use it for gameplay, one for example allows you to shoot at enemies with your gaze.
Not sure, but it would be cool to do something that didn’t require cameras, like EMG to detect eye motion from muscle activation. It would be hard to get the necessary accuracy though.
I don’t know about the Tobii glasses, but I have a Tobii head/eye tracker that attaches to my monitor, and it works incredibly well and was only $200 or so. I’d be surprised if this isn’t an essentially solved problem at this point.
So, guy who has deployed eye-scanning machines all over Africa and has found that many of them have been hacked and are giving incorrect responses suddenly has a friend with ALS and is willing to fund better quality eye tracking?
Either:
- Part of the whole world-coin thing was privately trying to get the data to help his friend
- He doesn't want to say "looking to develop eye tracking tech for my world-coin scam", since most devs won't touch that thing. Conveniently found a "friend" with ALS.
Saying, on behalf of a friend, that he doesn't believe PG.
I believe you may be mixing up Paul Graham and Sam Altman. Easy to do because they both ran YC at different times, but WorldCoin is from Sam Altman and this tweet is from PG.
It's not 'easy' to mix up people who you are accusing of lying about having a friend with ALS for profit.
Worldcoin also has nothing to do with eye-tracking, unless you purposefully play with English words. The request is optimizing software & eye tracking. Eye tracking itself is a solved problem, maybe more is needed for disabilities but that is not Worldcoin.
And since the request is for the Bay Area to meet the friend with ALS, what was the game plan here, invite them in and trap them in his lair instead?
I dunno, they seem like pretty different domains to me, despite both looking at the eye. Generating an image of the eye that's consistent across different scans in different environments but different from person to person, is one challenging problem. Figuring out where two pupils are in tandem while handling saccades and different focal depths is a different challenging problem. The former is a detailed static view of an eye. The latter is a less detailed picture of the eye itself (probably pretty easy to look at an eye and identify the pupil) but a complex dynamic algorithmic problem.
He currently has to put up with fakers, dupes that are not picked up, etc.
Once he can require you to focus on a series of letters with your eyes, it will be harder to fake, easier to pick up dupes. Aside for the fact that getting a "typing cadence" with each individual's eyes is probably valuable in and of itself.
I've been working on the menuing side [1] based on crossing Fitt's Law with Huffman trees. But, don't know the constraints for ALS.
Hopefully, whomever takes this on doesn't take the standard Accessibility approach, which is adding an extra layer of complexity on an existing UI.
A good friend, Gordon Fuller, found out he was going blind. So, he co-founded one of the first VR startups in the 90's. Why? For wayfinding.
What we came up with is a concept of Universal design. Start over from first principles. Seeing Gordon use an Accessible UI is painful to watch, it takes three times as many steps to navigate and confirm. So, what is the factor? 0.3 X?
Imagine if we could refactor all apps with a LLM, and then couple it with an auto compete menu. Within that menu is personal history of all your past transversals.
What would be the result? A 10X? Would my sister in a wheelchair be able to use it? Would love to find out!
Another route might also be sub-vocalization[1], like TTS for your thoughts. I recently picked up some cheap toys to get started trying to emulate the results[2].
I agree! My PhD thesis is on this topic [1]. We’ve also done a very limited pilot test on a patient with ALS, with above random chance. Actual results may vary heavily on individual disease progression—the more motor recruitment that’s intact, the better.
I guess I'm just interested in more information. Do people who aren't NASA have this working? I am just learning this sci-fi feature from Ender's universe is possibly a reality, and if I can check it out, I want to.
TBH, other than the NASA resources, I'm building it from scratch. Goal would be to record the signal while prompting a user to say/think a subset of words. Afterwards the HOPE is that I will be able to train a simple- functional- model based on those inputs.
Is the lack of mentioning Apple deliberate? It seems like they've already poured a lot of R&D into this for the Vision Pro, which might be exactly the kind of thing the friend needs.
I used to work on it and have spent tons of time in the headset. The eye tracking is next-level and it's really the only platform that exists with eye tracking as a primary input method. I'm pretty confident it will greatly improve your friend's quality of life.
Because of that, I'm also sure that eye tracking will go mainstream in other areas once the Vision Pro is released once everyone else catches on to it as a great input method.
This is pretty much exactly why I vehemently disagree with Apple's decision to draw such a firm line in the sand preventing devs from accessing the eye/gaze data directly. I'm part of an academic spin-off start-up that specializes in analyzing gaze and movement data. Locking the gaze information outside of the app sandbox severely hampers the ability to quickly iterate design and UI patterns that could be game changing for accessibility. Hopefully they make accommodations moving forward for these circumstances.
The issue is doubly close to my heart because my father has ALS and is nearly at the point where eye-tracking will be his only means of communicating effectively with the world. While existing Tobii systems work well enough, typing with your eyes is still exhausting to do.
Ultimately I don't think a platform like the vision pro is suitable for ALS patients, especially later term. They cannot support the weight of the headset and/or fatigue will set in rapidly. Many (including my father) also require use of a ventilator, accompanied with a mask that can seal effectively enough to support the positive pressure necessary to inflate their lungs. Unless the form factor for HMD's minimalizes significantly, it will likely interfere with the respirator's efficacy.
I don’t know much about those medical conditions, but it doesn’t take that much imagination to understand that access to eye-gaze data would pretty much give developers mind-reading abilities against whoever is wearing the headset. As the platform matures it will probably be a whole discussion around how it works, who gets access to it and for what reason. I could imagine Apple putting their weight behind developing all sorts of wild disability features.
Developers control what goes in front of the user and where, we'll still be able to tell plenty about a user's decision making process given that and how their head and hands navigate the space. There are plenty of companies that specialize in this as their entire product offering, assessing fitness for duty, alertness, attention mapping, etc. Plenty of published research on the matter as well.
The supposed security of blackboxing the eye data itself is illusory and functionally just for marketing.
The level of eye tracking performance for general population interactions is really only possible when you control the illumination like in a VR headset. A Vision Pro might work for the friend in question. More generally this requires the full vr display to make it work. See through AR or just plane glasses will not be nearly as good, and I think that will cap the general acceptance.
Is it doing something fundamentally different from what everyone else is doing (infrared light source, do some flavour of pupil segmentation and pose estimation)?
Is the eye-tracking performance/accuracy step change on Apple's headset purely just a software/algo change? Or is it actually using a new principle/apparatus for eye-tracking?
I don't think there's anything revolutionary, just a lot of parts working very well in tandem:
- Multiple cameras per eye, and at a very short distance from your eye
- The screen is fixed relative to the cameras for all devices, there's no worry about that half of the equation getting off calibration or differing for every customer
- OS is built around eye tracking, which means there won't be any actions that are unnaturally hard to perform with eye tracking
Doesn't it rely on external cameras to see users hands and use that as the "click" inputs? Seems like that negates usage for ALS cases.
Also, I'm not an ALS expert, but if the only muscular control is in the eyes, then lack of control in the head/neck probably breaks some assumptions about how the vision headset works (just a guess though).
It does not require using one's hands to click, it supports various input hardware (keyboard, mouse, switch, etc.). If someone has control of basically any muscle, it can use a switch input. The Vision Pro also has Dwell Control, activating things by keeping your gaze on it long enough, but I don't know whether it can currently be solely operated using nothing but one's eyes.
Not just multiple companies, multiple approaches. Eye tracking is exhausting, and that's pretty fundamental to the modality - dwell time requires significant control to be usable as a click, and even able bodied people find it exhausting. Some folks have tried doing eye tracking and using something else (EMG for example) as the click, but it doesn't work consistently for the population. ALS is also progressive, and people lose their eye control. Blackrock Neurotech has been working on a brain implant, with spinal cord injury as a first target population (because they're less fragile, among other reasons), and it works for current research patients, but medical devices take a lot of time, money and work to get cleared in the US. The implant itself is cleared, but the FDA wants the entire system to be cleared too.
Agreed. It would be great if the hardware was more affordable/accessible as well. That's potentially a barrier to entry worth addressing for devs who might otherwise be interested in tackling the problem, but don't have the quality eye tracking hardware to start. A Tobii-like hardware devkit could be a starting point.
1) its eye tracking isn't good enough for this kind of application.
2) direct access to the gaze vector is disabled
3) its really intrusive
4) its heavy.
5) it doesn't exist(in consumer world) yet.
The goal is to enable someone who has motor control issues, be able to communicate directly with the outside world. Shoving a heavy skimask that totally obscures the outside world on their face directly stops that.
Not only that, but you'll need to create and keep up to date the software needed to make a communicator. Apple are many thing, but it's new platforms are not stable, rapid os updates will break things.
> A friend of mine has ALS and can only move his eyes. He has an eye-controlled keyboard, but it's not very good. Can you make him a better one?
When I worked for one of the big game engines I got contacted by the makers of the tech that Stephen Hawking used to communicate, which includes an eye tracker:
I would love to hear pg's analysis of the business case for this company.
By my math, 5k people in the US are diagnosed per year, and if your keyboard costs $1k, then your ARR is $5m, and maybe the company valuation is $50m. Numerically, this is pretty far from the goal of a typical YC company.
I hate to be so cold-hearted about the calculations, but I've had a few friends get really passionate about assistive tech, and then get crushed by the financial realities. Just from the comments, you can see how many startups went either the military route or got acquired into VR programs.
The worst I've seen, btw, is trying to build a better powered wheelchair. All the tech is out there to make powered wheelchairs less bulky and more functional, but the costs of getting it approved for health insurance to pay the price, combined with any possible risk of them falling over, combined with the tiny market you are addressing makes it nearly impossible to develop and ship an improvement. I do hope that we reach a tipping point in the near future where a new wheelchair makes sense to build, because something more nimble would be a big improvement to people's lives.
ALS is a configuration. Blindness is a configuration. Conventional UIs are built for a different configuration. Accesible UI usually overloads the conventional. If you start from scatch, sometimes you find something new. If you're lucky, the new thing scales to other configurations.
For example, I wrote a NLP parser for a calendar app, at Tempo.AI. It was much more efficient than the visual interface. And thus, it was accessible. But, it didn't use the accessible idiom. Instead, it was universally more efficient, whether you are blind or not.
A good example is a wheelchair accessible doorway. One method is to have a button at wheelchair height. The other method is to have the door open with an electronic eye. The first is Accessible. The second is Universal. Doesn't matter if you are in a wheelchair or not. It's a throughput multiplier.
Another use case that I've seen talked about when it comes to eye-tracking technology is as it relates to the ad industry (it's always the freakin ad industry...). Eye-tracking is another way to detect what you're paying attention to. Currently, a lot of the ad business depends on measuring impressions, but they do so with heuristics. Eye-tracking can actually tell you if a user is looking at something vs pretending they do.
this is undoubtedly a direction Facebook is going with Oculus' eye tracking VR headset: serve arbitrary ads or other content, record subconscious saccades, build psychographic profile of unknowing and unwilling victim
Before hearing the business case, it would be good to know what is already available and has been found wanting. There seem to be a lot of people saying that they have working systems of some kind.
I have absolutely no idea of the state of play of eye-based input devices, but I wouldn't expect an entire eye-input keyboard system to be anywhere near as cheap as $1K, but rather a figure 20-100x that.
As someone who suffered some severe mobility impairment a few years ago and relied extensively on eye tracking for just over a year, https://precisiongazemouse.org/ (Windows) and https://talonvoice.com/ (multiplatform) are great. In my experience the hardware is already surprisingly good, in that you get accuracy to within an inch or half an inch depending on your training. Rather, it's all about the UX wrapped around it, as a few other comments have raised.
IMO Talon wins* for that by supporting voice recognition and mouth noises (think lip popping), which are less fatiguing than one-eye blinks for common actions like clicking. The creator is active here sometimes.
I remember seeing a program years ago, which used the mouse cursor in a really neat way to enter text. Seems like it would be far better than clicking on keys of a virtual keyboard, but I can't remember the name of this program nor seem to find it...
Will probably get some of this wrong, but just in case it rings a bell (or someone wants to reinvent it - wouldn't be hard):
The interface felt like a side-scrolling through through a map of characters. Moving left and right controlled speed through the characters; for instance moving to the left extent would backspace, and moving further to the right would enter more characters per time.
Up and down would select the next character - in my memory these are presented as a stack of map-coloured boxes where each box held a letter (or, group of letters?), say 'a' to 'z' top-to-bottom, plus a few punctuation marks. The height of each box was proportional to the likelihood that letter would be the next you'd want, so the most likely targets would be easier+quicker to navigate to. Navigating in to a box for a character would "type" it. IIRC, at any instant, you could see a couple levels of letters, so if you had entered c-o, maybe 'o' and 'u' would be particularly large, and inside the 'o' box you might see that 'l' and 'k' are bigger so it's easy to write "cool" or "cook".
(I do hardware+firmware in Rust and regularly reference Richard Hamming, Fred Brooks, Donald Norman, Tufte. Could be up for a change)
Geometrically 50% of the movement does nothing (going left is a null action). In terms of distance moved by the pointer it would be more efficient to arrange the options as a circle around the pointer, with maybe the bottom 10% of the circle used for the null action
A possible benefit is 90% of the screen could then be used to display a larger set of options, which could let the user bisect their choices more quickly
That is an interesting idea! The space of possible next characters expands polynomially, so as you say, the display might be better used with a polar grid compared to cartesian.
I'm also wondering if we might actually have run into each other when you lived in Christchurch around the quake times (as I did) because I do remember meeting someone who did Antarctic neutrino-related things at one of the regular local end-of-the-week tech gatherings... If so, it'd be an even smaller world. :)
Finally, I don't suppose you've got another of those Rust + firmware + NZ remote jobs/contracts lying around do you? :D I'm currently "exploring available work opportunities" but have almost given up on finding a role with even one of that trifecta let alone the whole bunch!
If you do decide to be "up for a change" with the eye-tracking thing feel free to get in touch if you're looking for someone to continue your current work. :D
Anyway, thanks for reading all my verbiage & I hope at least some of it was of interest. :)
Okay, so, your comment about a "Dasher + Guitar Hero music theory/improvisation practice program" just sent me down a huge rabbit hole...
Well, rabbit hole(s) plural, I guess, most not directly related. :D
Largely because I made the "mistake" of looking at your HN profile & discovering you're also in NZ & we seem to have somewhat overlapping interests (and an affinity for "bacon" in account names, apparently), so, some thoughts[0]... :)
Not really what I was anticipating finding but seems relevant to your interests--I don't think it was already in your resource list.
While the thesis addresses the Renesas 78K0 rather than the Renesas 78K0R, from a brief look at the "Flash Protection" PDF Application Note in your resource list it seems there's a large overlap.
Perhaps most significantly the author presents "novel methods" that combine bootloader binary analysis with constraint-based power glitching in an effort to improve on the results described in "Shaping the Glitch".
While I haven't read the entire 186 pages :D they theorize that using their approach extracting 8kB firmware might only take ~10 hours.
So, an interesting adjacent read even if it turns out not to be directly applicable to your situation.
Given I have an interest in & a little experience with firmware reversing my original thought was to maybe provide some hopefully helpful references that more generically related to firmware reversing but more specific is good too, I guess. :)
On the CAN tooling/info front, you might be interested in taking a look at my "Adequate CAN" list which I originally wrote-up for a client a couple years ago: https://gitlab.com/RancidBacon/adequate-can
In terms of how to approach RE, other than just "getting started & digging in" & learning by doing, I've sometimes found it informative to read other people's firmware reverse engineering write-ups to learn about potentially useful approaches/tools.
Anyway, hopefully some of this is helpful!
[0] I have a tendency to be a little... "verbose" and/or "thorough" (depending on one's POV :) ) so I'll probably split this over a couple of comments, in case I run out of steam while writing and for topic separation.
# Topic 2: Dasher + Guitar Hero style music theory/improvisation practice program
Back "on topic": I remember being quite enamoured/fascinated by dasher when I first encountered it. It's quite a unique interaction paradigm with the constant "forward movement" and "intelligent" pre-filtering/constraint of options with size-based prioritization.
Your suggestion to extend this interaction style for use in the music theory domain immediately appealed to me, as it intersects with some musical things I've been thinking about/playing with recently.
Over the past couple of years I've been playing around with ("rules based" rather than ML) procedural music generation primarily in the context of games.
This has been motivated by a couple of things: partly a procgen project is helpful as a driver for gaining an deeper understanding of music theory which I would like to develop for my own composition/production; and, I'm really interested in exploring ways of providing people with the experience of actually composing/creating their own music--which is something I think many people perceive as something only "musicians"/"composers" can do.
The latter is driven by my own music composition/creation/education experience: I learned piano as a kid for about a year until it was "mutually agreed" that if I wasn't going to practice perhaps it would be best to stop. :D But I've always really enjoyed music, particularly electronic/dance/EDM, and wanted to also create it & not just consume it--over the years I played around a tiny amount with creating some but gravitated toward DJing as my primary means of musical expression.
Then a few years ago I started "more seriously" creating tracks with LMMS (a FLOSS DAW https://lmms.io) and while progress was slow it was still nice to be able to enjoy the results.
But I grew frustrated/dissatisfied by the fact that I didn't really know how to add more of a melodic component to my music. (I'm an Anthemic Trance guy from way back. :D )
Over a couple of years after butting my head up against Music Theory a few times and bouncing off again (not unlike my experience with Rust :D ) one day I suddenly somehow "saw" some of the (simplified) Music Theory patterns/rules that I'd not internalised/understood previously.
And then I could add melody to my tracks! :o I mean they weren't masterpieces but it sounded like music! It blew my mind. :)
Not long after I realised something I found quite profound: it felt like music, instrument skills & music theory had only ever been presented to me as a thing that you did so you could playother people's music but I never wanted to play other people's music, I wanted to create my own!
Which then triggered a period of "Why didn't anyone teach me years ago when I was a kid that you could create your own music by starting with a few simple rules & building on them? Here I was "many" years later voluntarily learning about music theory, trying to apply it and even practising scales! :o
Anyway, that experience made me wonder if other people have experienced music & its creation in the same way and what opportunities there might be (particularly within a game/casual context) to provide those people with their first taste of creating music through a "guided" experience of just playing (in both senses of the word).
So, yeah, the "Guitar Dasher"/"Piano Dasher" concept aligns quite nicely with that. :)
Not that anyone asked me. :D
Couple of related things:
* Your suggestion also reminded me of another FLOSS DAW I played around with called Helio which has a "chord tool" (https://docs.helio.fm/tips-and-tricks.html#chord-tool) which appears as a pie-menu pre-populated with chords that fit with the current scale/root. I seem to recall that there are commercial DAWs that also have a similar UI.
* While I'm not particularly happy with its current state (really need to upload the most recent version of the code, which I'm fractionally happier with) here's my first foray into music procgen for a game jam entry (with a "debug quality UI" for controlling the output), if you're interested in checking it out: https://rancidbacon.itch.io/the-conductor
* And from a different angle here's another game jam entry where the concept I was playing with was essentially using music theory concepts as the basis for creating combat interactions/patterns (e.g. "Oh, no, how am I going to harmonize with whatever that was that the boss just played?!") and it all takes place on the "Grand Staff"/"Great Stave": https://rancidbacon.itch.io/stave-off
(Unfortunately as often seems to be the case I ended up spending more time fighting with a Unicode music engraving font/standard than I did writing game play for that last one. :) )
Huh. I wrote a paper for my undergraduate dissertation on eye tracking using a laptop camera, and it ended up published and I won a scholarship award (for €150, imagine that). I wonder if it's time to dust off that project
I agree, eye tracking is going to have really broad applications. I've been interested in eye tracking for over a decade, and in fact built my own eye tracker, joined a startup, and got acquired by Google[1]. But there's way more to do. We've barely scratched the surface of what's possible with eye tracking and I'd love to take a second crack at it.
The "Daydream View" VR headset developed before we were acquired totally bombed. Then a bunch of VR tech, including both long running internal projects and recent acquisitions, got shelved.
Adhawk, adhawk.io, has the only all day ultralight eye tracking wearable I'm aware of, all MEMS based with ultra high scan rates, 500Hz+ and research grade accuracy. For ALS u likely need something light and frictionless, wearing a hot and heavy headset all day probably doesn't work.
I suspect the real moneymakers for such startups have very little to do with ALS. ALS demand is fortunately small, and can't lead to VC desired growth curve. Imagine instead using it in a classroom to ensure the kids pay attention. Or making sure you see the advertisement.
>>Imagine instead using it in a classroom to ensure the kids pay attention.
>Imagine being a parent and being ok with this?
I'm sure there are many tiger-style parents which would be perfectly ok, nay, thrilled!, with this.
---
I didn't describe a utopian scenario. More like a dystopian one. The Defense applications do exist, but I expect the advertisers to dominate as they usually do.
Yes, the ALS/disability angle is noble. Viewed another way, the entire human race is afflicted by the disability of not having access to eye-tracking (and other) technologies. Paul Graham and co. are also invested in companies that are going to be highly enabled and boosted by the growth of eye-tracking and related technologies. I don't view his statement of motivation related to ALS as insincere, I just also notice that it's accessible, easily understandable, and also in line with other aspects of Paul's motivation (and that's a good thing).
I would also recommend Jean-Dominique Bauby's Le Scaphandre et le Papillon to anyone interested in this topic. Typing using eye movements was used in that book in a slow, inefficient manner. In the book's case, the question one should ask is, was his UI paced at the exact correct speed? I was and still am deeply emotionally moved by what the author was able to accomplish and convey. I am unsure if a faster keyboard would have made a meaningful and positive difference in that particular case, to the author's quality of life. I'll need to give that book another read with that question in mind.
Happily, I expect eye tracking to find fascinating, novel and unexpected applications. As others have stated, UI/UX design is an interesting part of this puzzle. For example, if you ask an LLM to output short branches of text and have a writer look at the words that he wants to convey. It's definitely blurring the line between reading and writing. Myself, finding writing to be a tactile exercise, I think that emotional state comes into play. That's what I'm interested in. Yes, can you literally read someone's eyes and tell what they are thinking?
I'm consulting with an Australian group called Control Bionics. They have a US company & office, with CTO and sales team in Ohio, but software engineering is done in AU. Their primary product is an electromyography and accelerometer hardware device to detect muscle activations or movements, and then most commonly used as a mouse-click substitute in conjunction with third-party eye-gaze hardware proving the cursor. (I've also designed them an autonomous wheelchair module, but that's another story...)
@pg - If your friend has not tried adding a mouse-click via something they can activate other than eye-gaze, this would be worth a shot. We have a lot of MND patients who use our combination to great success. If they can twitch an eyebrow, wiggle a toe or a finger, or even flex their abdomen, we can put electrodes there and give them a way forward.
Also, my contact details are in my profile. I'd be happy to put you in touch with our CEO and I'm confident that offers of funding would be of interest. The company is listed on the Australian stock exchange, but could likely go much further with a direct injection of capital to bolster the engineering team.
Eye tracking is essentially a model of visual attention. Visual attention is part of the overall attention space and big companies and use-cases are built around visual attention. Today we track attention by explicit interactions, if we can model around implicitly observable interactions - then we have a much larger observable data space around the user.,
I did a fun project a few years ago with eye tracking.
We built a prototype for roadside sobriety checks. The idea was to take race/subjectivity out of the equation in these traffic stops.
We modified an oculus quest and added IR LEDs and cameras with small PI zero's. I wrote software for the quest that gave instructions and had a series of examinations where you'd follow a 3D ball, the screen would brighten and darken, and several others while I looked for eye jerks (saccades) and pupil dilation. The officer was able to see your pupil (enlarged) on a laptop in real time and we'd mark suspicious times on the video timeline for review.
It was an interesting combination of video decoding, OpenCV and real-time streams with a pretty slick UI. The Pi Zero was easily capable of handling real-time video stream decoding, OpenCV and Node. Where I ran into performance problems I wrote node -> c++ bindings.
We did it all on something silly like a 50k budget. Neat project.
What do you think are some challenges that an eyetracker in this specific context has to face? What is your friend mostly struggling with the current solutions? Are there tracking specific challenges related to ALS? Is it mostly a UI/"better prediction" interface issue?
With my group we are developing an eyetracker for studying developmental and clinical populations, which typically present challenges to conventional eyetrackers. It is a spin off from our academic work with infants, and we already have a study almost done that uses it. We are still into the very beginning phase in terms of where this may lead us, but we are interested in looking into contexts where eyetracking for different reasons may be more challenging.
PG mention that the solution his friend used wasn't any good. How does the best system there is out today work? And what different solutions are there?
How about a library that starts loading a link when you look at it with intent. Or maybe with BCI integration that detects the moment you decide you want to access it.
Or how about a UI that automatically adapts to your eye movement and access patterns to minimize the amount of eye movement required to complete your most common tasks by rearranging the UI elements.
I thought this was solved long time ago, I wrote a program many years ago using kinect that tracks the center of the eye pretty precisely, using color gradients. The pupil is pretty uniform in every human being (it's black) surrounded by some color and then white. Even just a few pixels are enough to do it.
That's an interesting idea. How do you see it being beneficial to machine learning models, other than (I assume) it could work more efficiently within less foveal regions? Perhaps cases where you want the vision to emulate human vision?
Running a fast network on sparse data then calling one optimized for a task on a subset seems like a good optimization and dealing with video we are probably going to need them.
After you parse what an object is, tracking it doesn't take anywhere near the effort of original segmentation. No need to re-evaluate until something changes.
Maybe even use activations to turn on and off networks. "Oh text better load ocr into memory"
And it does inform a lot of our built world
It's strange to think that when watching a movie only 10% is in focus.
Eye movement does provide a lot of information to other people and I think the physical movement produces feedback for velocities and things too. Mimicking biology is often a good bet.
I would like to look at the problem more deeply, the eyes can be tracked but what about facial movement, the more data the better training for machine learning
It is easy to detect what people are looking at. Solved problem. It’s difficult to effectively use such a low—bandwidth signal, and that’s where LLM come in. You need an LLM to anticipate what the user wants to say. An LLM fine-tuned on the user’s needs may have sufficiently low entropy that the small number of bits/s you’re getting from the eye-tracker can reliably identify the intended message.
I believe Paul Graham can Google or use AI and already knows about the companies and links you posted. His post was a call to action to connect with people working on yet to be discovered innovations and inspire those and others quietly working to come forward and connect with him.
Aw, that's nice of pg to want something better for his friend. As cynical as we are about technology, new developments can be so fantastic for accessibility and better quality of life.
And if that were where the story ended we'd have an honest feel-good going. New developments -could- be fantastic for accessibility and QoL, but without exception they just end up getting sucked into a marketing surveillance suite.
Lithium extraction is also very bad for the environment. Requires insane amounts of water, requires often times destroying forests or other natural areas that are on top of lithium, etc.
Sure I use lithium ion batteries too, in the developed country I live in, I can't go and live in a cave. But I won't buy a SUV that weighs 2 tons to sit alone in it most of the time and have it parked 95% of the time in a garage.
The "what have you done" attack is a bit too easy and, really, this is not about me.
We should strive to be sober, to buy repairable goods whenever possible, to educate ourselves on environmental impact of things, and perhaps more importantly to teach this to our children.
This thread is about eye tracking. You went from talking about the congo for some reason to now shifting to "lithium takes a lot of water".
But I won't buy a SUV that weighs 2 tons to sit alone in it most of the time and have it parked 95% of the time in a garage.
What are you even talking about? This whole thread is about eye tracking.
We should strive to be sober, to buy repairable goods whenever possible, to educate ourselves on environmental impact of things, and perhaps more importantly to teach this to our children
It's all related. High tech has a lot of negative impacts which are rarely if ever evaluated and more importantly compensated. The reason is they bring in money.
The entire supply chain and lifecycle of a product should be evaluated to understand if it is a net positive or a net negative. Not only the finished product: "wow eye tracking how cool let's deploy everywhere regardless of how the materials for that tech are mined and how the device will be recycled once obsolete".
To be clear, in your mind when someone talks about wanting to fund certain types of software, it is someone's duty to talk about lithium batteries using lots of water?
It seems you want to keep asking the same questions over and over phrased differently and in a way that tries to discredit me or make me sound dumb, so forgive me but this discussion has now reached its end.
So you're saying there's a final frontier in the mapping of intention and the discernment of preferences ... and you'd like some bright young things to explore that possibility space under the guise of (unimpeachable good cause) ?
Go to hell.
Unless, of course, you'd like to commit the funded work to the free commons, unencumbered by patents and copyrights, and free to use by any entity for any purpose.
Im something of a conspiracy theorist myself, but this is irrational and unwarranted: if they wanted to research something strategic, they could easily summon the brain power (see OpenAI and the creation of near-general intelligence, and then ask where Facebooks $40B on metaverse smokescreen went to.)
PG is a good guy - he sides with entrepreneurs, and created an industry-wide standard for a founder friendly seed stage investment deal (called a SAFE) in a world of hugely predatory deal makers. And from his twitter its pretty clear that he's biased towards fairness and humanity in general.
Read some of PG's essays. People dont share deep insights and knowledge like that for free if they are made of the wrong stuff.
My comment violates the HN guideline to "respond to the strongest plausible interpretation of what someone says" and I beg the op's pardon.
A low (or no) return investment in basic technology to help the disabled better express their intentions - for their own private uses - is an unalloyed good.
Of my two interpretations, I will set aside my initial, very cynical reaction and wait to see what actually comes from this.
Only that this is already being done for "nefarious" (read advertising) purposes. You do not need any "good excuse" to do just that, just money. And, essentially, being in the commons or not is consequential in how it is to be used. There are a million of topics and cases one is warranted to be suspicious and argumentative against, than when talking about dealing with medical conditions.
Is anyone speedrunning this? I can't find anything with a quick look.
I'd consider an approach like the human powered helicopter parable.
I'd create a model for the eyes / face. So a screen like a phone is put in front of a camera with a model face and controllable eyes that you use software to control. Maybe skip the screen in front of a camera and link straight into the video feed.
It knows the limits of the eyes (different models for different people and situations) can measure fatigue etc.
A lot of past chronic diseases have had cures found. Leprosy and tuberculosis for example. ALS is a tricky one though, not well understood and quite likely prion like.
Someone should do this, but for the love of god, DO NOT take any venture capital to do it. No matter how well-intentioned the VCs are at the start, eventually your eye tracking startup will 100% be used for advertising as investors in your Series D need to make an exit or take you public.
As we are talking about a company to help people with medical problems.
Second Sight was giving patients artificial eyes. When they ran out of funding, they closed shop. The patients lost support system for their eyes. If anything goes wrong with their artificial eye, there is no one to repair or fix it. They just have to carry a piece of useless metal junk in their head.
Except Apple has stated they will encrypt that information and not supply it to apps to avoid targeting or fingerprinting. And second how would that help someone with ALS?
> And second how would that help someone with ALS?
They won't share information on where you look, but they will share info on where you 'click', which is used to navigate apps. IIRC you are supposed to use your fingers to do this, and other actions, but I imagine that Apple's accessibility team will have an alternate mode for people with motor limitations. It could be a long-blink, or rapid blinking, for example.
Also donning the AVP headset might not be possible for someone with advanced ALS. A fixed outside-in apparatus (probably attached to the patient’s wheelchair) would make more sense.
There are a bunch of hardware out there to get gaze vectors, the problem is that they rely on a generalised model of the eye. With fine tuning you can go from >5 degree of error, to <1-2
Not really a huge PG fan, but this is what billionaires should be doing: see where a need is exists and put some of your insane wealth towards making an improvement. This why I respect Elon even though I don’t really like him; he puts his money to use, in a very public manner.
Yes, and? There is no plausible near future where anyone else (except maybe his heirs) directs how that wealth is used. Better that it go towards socially useful investments than megayachts or land hoarding.
I'm not arguing that Elon doesn't put his money to use publicly, but I'm interested how you think he does. Do you mean he uses his wealth privately in a way that benefits the public, and/or he does it in public view?
I fail to see how Tesla fixes an "important problem". They're big electric SUV with huge batteries that require tons of mining and destroy the environment. They're largely non repairable. The full self driving feature seems to be vapor ware.
The space stuff looked interesting until he started sending all kinds of things into space to become an ISP, polluting space the same way Tesla pollutes the earth.
The real issues are still here: poverty, lack of access to clean water, etc. How does Musk fix that?
This had nothing to do with conservative talking points. It has to do with current capitalism always looking for money but letting the real societal issues ongoing because fixing them doesn't pay enough dividends.
Simplified, with current capitalism, doing the right thing doesn't pay off. We have more than enough know-how and technology to fix real issues but we'd rather put rich people in overweight electric cars.
Yes, full self driving could help save lifes. Space exploration helps us better understand our world. But often times, current capitalism encourages the use of technology and know-how for things that benefit a minority at the expense of the most vulnerable.
It’s not Musk’s responsibility to fix societal problems, it’s the government’s. He sucks as a person but the companies he’s in charge of are quite useful and I’m happy they exist. I’m watching my government fail every single day at addressing basic issues like homelessness despite near infinite money. They would fail at fast satellite internet and electric cars too.
I don’t entirely disagree with your point of view, I just question what the hell we pay taxes for if we (the people) are still responsible for all the societal issues. If I have to dedicate a portion of my day’s effort to chipping away at these things then why is the government entitled to a significant portion of my money? It doesn’t make sense to me.
YC was investing in ways the traditional VCs weren’t when it started, and coding HN was a part of it. I doubt I’m the only one who had having a few HN tech support emails from PG.
Hmm. I was curious and started looking for the comments you mention, and was pleasantly surprised to find zero pg hate comments, downvoted or otherwise. People seem pretty on board with billionaires solving problems for disabled people in exchange for mutual upside.
I used to be bothered by those kinds of sentiments too, by the way. The way I got over it was to realize how many people are just bitter, and not because of pg or YC. This is different than having an actual issue with pg or YC — it’s random noise rather than points worth listening to.
Weirder than the haters are the people who reply to his tweets. Some of them post bizarre things. I find it fascinating when people project their own feelings on him, whether it’s hate, admiration, or (my favorite, having been a victim of it myself) misplaced ambition.
I have a new approach of doing ML, where autodiff is replaced with something better. Magically a lot of things fall into place. This approach should make problems like this relatively straight forward.
If you can show your new approach works, sure. Usually this is done via papers in ML conferences, but if you have reproducible results on Github I'll take a look.
