computer vision

114 Articles

Smart Garbage Trucks Help With Street Maintenance

August 24, 2023 by 42 Comments

If you’ve ever had trouble with a footpath, bus stop, or other piece of urban infrastructure, you probably know the hassles of dealing with a local council. It can be incredibly difficult just to track down the right avenue to report issues, let alone get them sorted in a timely fashion.

In the suburban streets of one Australian city, though, that’s changing somewhat. New smart garbage trucks are becoming instruments of infrastructure surveillance, serving a dual purpose that could reshape urban management. Naturally, though, this new technology raises issues around ethics and privacy.

Continue reading “Smart Garbage Trucks Help With Street Maintenance” →

Several video clips of a robot arm manipulating objects in a kitchen environment, demonstrating some of the 12 generalized skills

RoboAgent Gets Its MT-ACT Together

August 20, 2023 by 6 Comments

Researchers at Carnegie Mellon University have shared a pre-print paper on generalized robot training within a small “practical data budget.” The team developed a system that breaks movement tasks into 12 “skills” (e.g., pick, place, slide, wipe) that can be combined to create new and complex trajectories within at least somewhat novel scenarios, called MT-ACT: Multi-Task Action Chunking Transformer. The authors write:

Trained merely on 7500 trajectories, we are demonstrating a universal RoboAgent that can exhibit a diverse set of 12 non-trivial manipulation skills (beyond picking/pushing, including articulated object manipulation and object re-orientation) across 38 tasks and can generalize them to 100s of diverse unseen scenarios (involving unseen objects, unseen tasks, and to completely unseen kitchens). RoboAgent can also evolve its capabilities with new experiences.

Continue reading “RoboAgent Gets Its MT-ACT Together” →

Spy Tech: Unshredding Documents

June 12, 2023 by 38 Comments

Bureaucracies generate paper, usually lots of paper. Anything you consider private — especially anything that could get you in trouble — should go in a “burn box” which is usually a locked trash can that is periodically emptied into an incinerator. However, what about a paper shredder? Who hasn’t seen a movie or TV show where the office furiously shreds papers as the FBI, SEC, or some other three-letter-agency is trying to crash the door down?

That might have been the scene in the late 1980s when Germany reunified. The East German Ministry of State Security — known as the Stasi — had records of unlawful activity and, probably, information about people of interest. The staff made a best effort to destroy these records, but they did not quite complete their task.

The collapsing East German government ordered documents destroyed, and many were pulped or burned. However, many of the documents were shredded by hand, stuffed into bags, and were awaiting final destruction. There were also some documents destroyed by the interim government in 1990. Today there are about 16,000 of these bags remaining, each with 2,500 to 3,000 pieces of pages in them.

Machine-shredded documents were too small to recover, but the hand-shredded documents should be possible to reconstruct. After all, they do it all the time in spy movies, right? With modern computers and vision systems, it should be a snap.

You’d think so, anyway.

Continue reading “Spy Tech: Unshredding Documents” →

Self-Driving Library For Python

May 16, 2023 by 3 Comments

Fully autonomous vehicles seem to perennially be just a few years away, sort of like the automotive equivalent of fusion power. But just because robotic vehicles haven’t made much progress on our roadways doesn’t mean we can’t play with the technology at the hobbyist level. You can embark on your own experimentation right now with this open source self-driving Python library.

Granted, this is a library built for much smaller vehicles, but it’s still quite full-featured. Known as Donkey Car, it’s mostly intended for what would otherwise be remote-controlled cars or robotics platforms. The library is built to be as minimalist as possible with modularity as a design principle, and includes the ability to self-drive with computer vision using machine-learning algorithms. It is capable of logging sensor data and interfacing with various controllers as well, either physical devices or through something like a browser.

To build a complete platform costs around $250 in parts, but most things needed for a Donkey Car compatible build are easily sourced and it won’t be too long before your own RC vehicle has more “full self-driving” capabilities than a Tesla, and potentially less risk of having a major security vulnerability as well.

Modern Dance Or Full-Body Keyboard? Why Not Both!

April 1, 2023 by 15 Comments

If you felt in your heart that Hackaday was a place that would forever be free from projects that require extensive choreography to pull off, we’re sorry to disappoint you. Because you’re going to need a level of coordination and gross motor skills that most of us probably lack if you’re going to type with this full-body, semaphore-powered keyboard.

This is another one of [Fletcher Heisler]’s alternative inputs projects, in the vein of his face-operated coding keyboard. The idea there was to be able to code with facial gestures while cradling a sleeping baby; this project is quite a bit more expressive. Pretty much all you need to know about the technical side of the project can be gleaned from the brilliant “Hello world!” segment at the start of the video below. [Fletcher] uses OpenCV and MediaPipe’s Pose library for pose estimation to decode the classic flag semaphore alphabet, which encodes characters in the angle of the signaler’s extended arms relative to their body. To extend the character set, [Fletcher] added a squat gesture for numbers, and a shift function controlled by opening and closing the hands. The jazz-hands thing is just a bonus.

Honestly, the hack here is mostly a brain hack — learning a complex series of gestures and stringing them together fluidly isn’t easy. [Fletcher] used a few earworms to help him master the character set and tune his code; the inevitable Rickroll was quite artistic, and watching him nail the [Johnny Cash] song was strangely satisfying. We also thoroughly enjoyed the group number at the end. Ooga chaka FTW.

Continue reading “Modern Dance Or Full-Body Keyboard? Why Not Both!” →

Hands-On: NVIDIA Jetson Orin Nano Developer Kit

March 21, 2023 by 35 Comments

NVIDIA’s Jetson line of single-board computers are doing something different in a vast sea of relatively similar Linux SBCs. Designed for edge computing applications, such as a robot that needs to perform high-speed computer vision while out in the field, they provide exceptional performance in a board that’s of comparable size and weight to other SBCs on the market. The only difference, as you might expect, is that they tend to cost a lot more: the current top of the line Jetson AGX Orin Developer Kit is $1999 USD

Luckily for hackers and makers like us, NVIDIA realized they needed an affordable gateway into their ecosystem, so they introduced the $99 Jetson Nano in 2019. The product proved so popular that just a year later the company refreshed it with a streamlined carrier board that dropped the cost of the kit down to an incredible $59. Looking to expand on that success even further, today NVIDIA announced a new upmarket entry into the Nano family that lies somewhere in the middle.

While the $499 price tag of the Jetson Orin Nano Developer Kit may be a bit steep for hobbyists, there’s no question that you get a lot for your money. Capable of performing 40 trillion operations per second (TOPS), NVIDIA estimates the Orin Nano is a staggering 80X as powerful as the previous Nano. It’s a level of performance that, admittedly, not every Hackaday reader needs on their workbench. But the allure of a palm-sized supercomputer is very real, and anyone with an interest in experimenting with machine learning would do well to weigh (literally, and figuratively) the Orin Nano against a desktop computer with a comparable NVIDIA graphics card.

We were provided with one of the very first Jetson Orin Nano Developer Kits before their official unveiling during NVIDIA GTC (GPU Technology Conference), and I’ve spent the last few days getting up close and personal with the hardware and software. After coming to terms with the fact that this tiny board is considerably more powerful than the computer I’m currently writing this on, I’m left excited to see what the community can accomplish with the incredible performance offered by this pint-sized system.

Continue reading “Hands-On: NVIDIA Jetson Orin Nano Developer Kit” →

Webcam VR

Immersive Virtual Reality From The Humble Webcam

March 12, 2023 by 12 Comments

[Russ Maschmeyer] and Spatial Commerce Projects developed WonkaVision to demonstrate how 3D eye tracking from a single webcam can support rendering a graphical virtual reality (VR) display with realistic depth and space. Spatial Commerce Projects is a Shopify lab working to provide concepts, prototypes, and tools to explore the crossroads of spatial computing and commerce.

The graphical output provides a real sense of depth and three-dimensional space using an optical illusion that reacts to the viewer’s eye position. The eye position is used to render view-dependent images. The computer screen is made to feel like a window into a realistic 3D virtual space where objects beyond the window appear to have depth and objects before the window appear to project out into the space in front of the screen. The resulting experience is like a 3D view into a virtual space. The downside is that the experience only works for one viewer.

Eye tracking is performed using Google’s MediaPipe Iris library, which relies on the fact that the iris diameter of the human eye is almost exactly 11.7 mm for most humans. Computer vision algorithms in the library use this geometrical fact to efficiently locate and track human irises with high accuracy.

Generation of view-dependent images based on tracking a viewer’s eye position was inspired by a classic hack from Johnny Lee to create a VR display using a Wiimote. Hopefully, these eye-tracking approaches will continue to evolve and provide improved motion-responsive views into immersive virtual spaces.