artificial intelligence

97 Articles

Love AI, But Don’t Love It Too Much

December 7, 2022 by 51 Comments

The up-and-coming Wonder of the World in software and  information circles , and particularly in those circles who talk about them, is AI. Give a magic machine a lot of stuff, ask it a question, and it will give you a meaningful and useful answer. It will create art, write books, compose music, and generally Change The World As We Know It. All this is genuinely impressive stuff, as anyone who has played with DALL-E will tell you. But it’s important to think about what the technology can and can’t do that’s new so as to not become caught up in the hype, and in doing that I’m immediately drawn to a previous career of mine. Continue reading “Love AI, But Don’t Love It Too Much” →

AI simulated drone flight track

Human Vs. AI Drone Racing At The University Of Zurich

November 26, 2022 by 12 Comments

[Thomas Bitmatta] and two other champion drone pilots visited the Robotics and Perception Group at the University of Zurich. The human pilots accepting the challenge to race drones against Artificial Intelligence “pilots” from the UZH research group.

The human pilots took on two different types of AI challengers. The first type leverages 36 tracking cameras positioned above the flight arena. Each camera captures 400 frames per second of video. The AI-piloted drone is fitted with at least four tracking markers that can be identified in the captured video frames. The captured video is fed into a computer vision and navigation system that analyzes the video to compute flight commands. The flight commands are then transmitted to the drone over the same wireless control channel that would be used by a human pilot’s remote controller.

The second type of AI pilot utilizes an onboard camera and autonomous machine vision processing. The “vision drone” is designed to leverage visual perception from the camera with little or no assistance from external computational power.

Ultimately, the human pilots were victorious over both types AI pilots. The AI systems do not (yet) robustly accommodate unexpected deviation from optimal conditions. Small variations in operating conditions often lead to mistakes and fatal crashes for the AI pilots.

Both of the AI pilot systems utilize some of the latest research in machine learning and neural networking to learn how to fly a given track. The systems train for a track using a combination of simulated environments and real-world flight deployments. In their final hours together, the university research team invited the human pilots to set up a new course for a final race. In less than two hours, the AI system trained to fly the new course. In the resulting real-world flight of the AI drone, its performance was quite impressive and shows great promise for the future of autonomous flight. We’re betting on the bots before long.

Continue reading “Human Vs. AI Drone Racing At The University Of Zurich” →

Researching Factorio…For Science

October 19, 2022 by 15 Comments

Science has affirmatively answered a lot of questions that, looking back, could be seen as bizarre to have asked in the first place. Questions like “can this moldy cheese cure disease” or “can this rock perform math if we give it some electricity.”  Among the more recent of this list is the question of whether or not the video game Factorio, in which the player constructs an elaborate factory, can be used as the basis for other academic work. As [Kenneth Reid] discusses in this talk, it most certainly can.

If you haven’t played the game, it’s a sort of real-time strategy (RTS) game where the player gathers materials to construct a factory while defending it from enemies. On the surface it might seem similar to Age of Empires or Starcraft, but its complexity is taken to extremes not found in other RTS games. The complexity hides nuance, and [Kenneth] points out that it’s an excellent simulator to study real-world problems such as vehicle routing problems, decision making, artificial intelligence, bin packing problems, and production planning, among a whole slew of other interesting areas of potential research.

[Kenneth] and his partners on this project also developed some software tools with interacting with a Factorio game without having to actually play it directly. The game includes an API which the team used to develop tools so that other researchers can use it as a basis for simulations and studies. There was a research paper published as well for more in-depth reading on the topic. We shouldn’t be too surprised that a game can be used in incredibly productive ways like this, either. Here’s another example of a toy being used to train engineers working in industrial automation.

Continue reading “Researching Factorio…For Science” →

Self-Driving Laboratories Do Research On Autopilot

September 29, 2022 by 16 Comments

Scientific research is a messy business. The road to learning new things and making discoveries is paved with hard labor, tough thinking, and plenty of dead ends. It’s a time-consuming, expensive endeavor, and for every success, there are thousands upon thousands of failures.

It’s a process so inefficient, you would think someone would have automated it already. The concept of the self-driving laboratory aims to do exactly that, and could revolutionize materials research in particular.

Continue reading “Self-Driving Laboratories Do Research On Autopilot” →

A 3D-Printed Nixie Clock Powered By An Arduino Runs This Robot

June 22, 2022 by 16 Comments

While it is hard to tell with a photo, this robot looks more like a model of an old- fashioned clock than anything resembling a Nixie tube. It’s the kind of project that could have been created by anyone with a little bit of Arduino tinkering experience. In this case, the 3D printer used by the Nixie clock project is a Prusa i3 (which is the same printer used to make the original Nixie tubes).

The Nixie clock project was started by a couple of students from the University of Washington who were bored one day and decided to have a go at creating their own timepiece. After a few prototypes and tinkering around with the code , they came up with a design for the clock that was more functional than ornate.

The result is a great example of how one can create a functional and aesthetically pleasing project with a little bit of free time.

Confused yet? You should be.

If you’ve read this far then you’re probably scratching your head and wondering what has come over Hackaday. Should you not have already guessed, the paragraphs above were generated by an AI — in this case Transformer — while the header image came by the popular DALL-E Mini, now rebranded as Craiyon. Both of them were given the most Hackaday title we could think of, “A 3D-Printed Nixie Clock Powered By An Arduino Runs This Robot“, and told to get on with it. This exercise was sparked by curiosity following the viral success of AI generators, which posed the question of whether an AI could make a passable stab at a Hackaday piece. Transformer runs on a prompt model in which the operator is given a choice of several sentence fragments so the text reflects those choices, but the act of choosing could equally have followed any of the options.

The text is both reassuring as a Hackaday writer because it doesn’t manage to convey anything useful, and also slightly shocking because from just that single prompt it’s created meaningful and clear sentences which on another day might have flowed from a Hackaday keyboard as part of a real article. It’s likely that we’ve found our way into whatever corpus trained its model and it’s also likely that subject matter so Hackaday-targeted would cause it to zero in on that part of its source material, but despite that it’s unnerving to realise that a computer somewhere might just have your number. For now though, Hackaday remains safe at the keyboards of a group of meatbags.

We’ve considered the potential for AI garbage before, when we looked at GitHub Copilot.

Researchers Build Neural Networks With Actual Neurons

March 1, 2022 by 21 Comments

Neural networks have become a hot topic over the last decade, put to work on jobs from recognizing image content to generating text and even playing video games. However, these artificial neural networks are essentially just piles of maths inside a computer, and while they are capable of great things, the technology hasn’t yet shown the capability to produce genuine intelligence.

Cortical Labs, based down in Melbourne, Australia, has a different approach. Rather than rely solely on silicon, their work involves growing real biological neurons on electrode arrays, allowing them to be interfaced with digital systems. Their latest work has shown promise that these real biological neural networks can be made to learn, according to a pre-print paper that is yet to go through peer review.
Continue reading “Researchers Build Neural Networks With Actual Neurons” →

E4 Empatica device for measuring location, temperature, skin conductance, sleep, etc. on arm

Wearable Sensor For Detecting Substance Use Disorder

December 9, 2021 by 14 Comments

Oftentimes, the feature set for our typical fitness-focused wearables feels a bit empty. Push notifications on your wrist? OK, fine. Counting your steps? Sure, why not. But how useful are those capabilities anyway? Well, what if wearables could be used for a more dignified purpose like helping people in recovery from substance use disorder (SUD)? That’s what the researchers at the University of Massachusetts Medical School aimed to find out.

In their paper, they used a wrist-worn wearable to measure locomotion, heart rate, skin temperature, and electrodermal activity of 38 SUD patients during their everyday lives. They wanted to detect periods of stress and craving, as these parameters are possible triggers of substance use. Furthermore, they had patients self-report times during the day when they felt stressed or had cravings, and used those reports to calibrate their model.

They tried a number of classification models such as decision trees, discriminant analysis, logistic regression, and others, but found the most success using support vector machines though they failed to discuss why they thought that was the case. In the end, they found that they could detect stress vs. non-stress with an accuracy of 81.3% and craving vs. no-craving with an accuracy of 82.1%. Not amazing accuracy, but given the dire need for medical advancements for SUD, it’s something to keep an eye on. Interestingly enough, they found that locomotion data alone had an accuracy of approximately 75% when it came to indicating stress and cravings.

Much ado has been made about the insufficient accuracy of wearable devices for medical diagnoses, particularly of those that measure activity and heart rate. Maybe their model would perform better, being trained on real-time measurements of cortisol, a more accurate physiological measure of stress.

Finally, what really stood out to us about this study was how willing patients were to use a wearable in their treatment strategy. It’s sad that society oftentimes has a very negative perception of SUD patients, leading to fewer treatment options for patients. But hopefully, with technological advancements such as this, we’re one step closer to a more equitable future of healthcare.