Grand Theft Auto V Used To Teach Self-Driving AI

September 16, 2016 by James Hobson 29 Comments

For all the complexity involved in driving, it becomes second nature to respond to pedestrians, environmental conditions, even the basic rules of the road. When it comes to AI, teaching machine learning algorithms how to drive in a virtual world makes sense when the real one is packed full of squishy humans and other potential catastrophes. So, why not use the wildly successful virtual world of Grand Theft Auto V to teach machine learning programs to operate a vehicle?

The hard problem with this approach is getting a large enough sample for the machine learning to be viable. The idea is this: the virtual world provides a far more efficient solution to supplying enough data to these programs compared to the time-consuming task of annotating object data from real-world images. In addition to scaling up the amount of data, researchers can manipulate weather, traffic, pedestrians and more to create complex conditions with which to train AI.

It’s pretty easy to teach the “rules of the road” — we do with 16-year-olds all the time. But those earliest drivers have already spent a lifetime observing the real world and watching parents drive. The virtual world inside GTA V is fantastically realistic. Humans are great pattern recognizers and fickle gamers would cry foul at anything that doesn’t analog real life. What we’re left with is a near-perfect source of test cases for machine learning to be applied to the hard part of self-drive: understanding the vastly variable world every vehicle encounters.

A team of researchers from Intel Labs and Darmstadt University in Germany created a program that automatically indexes the virtual world (as seen above), creating useful data for a machine learning program to consume. This isn’t a complete substitute for real-world experience mind you, but the freedom to make a few mistakes before putting an AI behind the wheel of a vehicle has the potential to speed up development of autonomous vehicles. Read the paper the team published Playing for Data: Ground Truth from Video Games.

Continue reading “Grand Theft Auto V Used To Teach Self-Driving AI” →

Robotic Arm From Cardboard

September 14, 2016 by Al Williams 7 Comments

Google showed the world that you could make a virtual reality headset from cardboard. We figure that might have been [Uladz] inspiration for creating a robotic arm also made out of cardboard. He says you can reproduce his design in about two hours.

You’ll need an Arduino and four hobby servo motors. The cardboard doesn’t weigh much, so you could probably use fairly small motors. In addition to the cardboard, there’s a piece of hardboard for the base and a few metal clips. You can control it all from the Arduino program or add an IR receiver if you want to run it by remote control. There’s a video of the arm–called CARDBIRD–in action, below.

Continue reading “Robotic Arm From Cardboard” →

Furuta Style Inverted Pendulum Is King Of Geek Desk Ornaments

September 14, 2016 by Gerrit Coetzee 19 Comments

Newton’s Cradle is thought of as the most elegant of executive desk toys. But that 20th-century dinosaur just got run off the road as [Ben Katz]’s Furuta pendulum streaks past in the fast lane, flipping the bird and heralding a new king of desk adornments.

This Furata pendulum has wonderfully smooth movement. You can watch it go through its dance in the video after the break. Obviously you agree that this is the desk objet d’art for the modern titan of industry (geek). Just don’t stop at watching it in action. The best part is the build log that [Ben] put together — this project has a little bit of everything!

Continue reading “Furuta Style Inverted Pendulum Is King Of Geek Desk Ornaments” →

Sewbo Robot Sews Up Automated Garment Manufacturing

September 12, 2016 by Moritz Walter 42 Comments

While robots enter other industries in herds, the assembly of garments has long been a tedious, human privilege. Now, for the first time, a robot has sewn an entire, wearable piece of garment. Sewbo, an industrial robot programmed to tackle the tricky task, assembles clothes and makes it look easy.

Continue reading “Sewbo Robot Sews Up Automated Garment Manufacturing” →

Super-small Robotic Joints Don’t Exist? They Do Now!

September 12, 2016 by Donald Papp 11 Comments

[Tim] needed very small, motorized joints for a robot. Unable to find anything to fit the bill, he designed his own tiny, robotic joints. Not only are these articulated and motorized, they are designed to be independent – each containing their own driver and microcontroller.

6mm geared motor next to LEGO [Source: Pololu]

None of the photos or video really give a good sense of just how small [Tim]’s design is. The motor (purple in the 3D render above, and pictured to the left) is a sub-micro planetary geared motor with a D shaped shaft. It is 6mm in diameter and 19mm long. One of these motors is almost entirely encapsulated within the screw it drives (green), forming a type of worm gear. As the motor turns the screw, a threaded ring moves up or down – which in turn moves the articulated shaft attached to the joint. A video is embedded below that shows the joint in action.

[Tim] originally tried 3D printing the pieces on his Lulzbot but it wasn’t up to the task. He’s currently using a Form 2 with white resin, which is able to make the tiny pieces just the way he needs them.

Continue reading “Super-small Robotic Joints Don’t Exist? They Do Now!” →

Hacklet 124 Running Robots And The Claw

September 10, 2016 by Adam Fabio 14 Comments

You never know what you’ll find when you open the projects feed on Hackaday.io. Most weeks, The Hacklet follows a theme of some sort. Sometimes I find projects that just look so cool that I have to get the word out about them.

Such is the case with this week’s first project, Mr. Runner created by [Alex Martin]. Mr. Runner is a quadruped robot that really looks the part. In fact, I’d say it looks like it’s ready to jump off the bench top. Like many of us, [Alex] has been inspired by Boston Dynamics, specifically their Wildcat robot. Wildcat had [Alex] searching the net for walking robot designs. He struck up something he liked with the work of [Dr. Fumiya Iida] and [Dr. Rolf Pfiefer]. In the mid 2000’s, the pair worked out of the University of Zurich. Mr. Runner is based upon their work, with plenty of design tweaks from [Alex].

The basic design is a quadruped with two servos per leg. The servos are at the body and the upper half of the leg. The knee and lower leg are connected by levers and a spring, forming something of a 4 bar linkage. The spring acts as a tendon, absorbing shock, and allowing energy from the servo to be stored and released while the robot runs. [Alex] is experimenting with gaits, controlled by a PC.

Mr. Runner wouldn’t be doing much running without a way to control those 8 servos. [Alex] started with an Arduino and a LynxMotion serial servo controller. This pairing served him well for the first generation of Mr. Runner. For the new version of the robot, he’s rolling his own board based upon Lynxmotion’s
BotBoarduino. The Gerber files have been sent off to OSH Park, and in about a week, Mr. Runner will be off to the races.

Another great recently updated project is Arcade Claw Game Claw Build by [Alex Anderson]. I spent way too many hours of my youth in arcades, and more than a few quarters went into claw games. Sure, they’re usually rigged, but who hasn’t been pulled in by the chance to test your skill and win a prize? A friend asked [Alex] to design an arcade style claw for a game. A seasoned CNC and 3D printing master, [Alex] grabbed his notebook and started sketching. Rack and pinion designs would work well, but didn’t within the constraints of the game. A leadscrew based design would also work, but would be two expensive. Finally, [Alex] settled on a design and fired up his CAD software. He started with two jaw systems to prove out the basic system. Once that was complete, [Alex] moved to a 4 jaw setup.

Much like the arcade games, the claw is actuated by a central plunger. The plunger drives linkages which move the 4 claw jaws. Everything looks good on paper, but when the CAD drawings meet the real world, things get complicated quickly. The initial design relied on a 3D printed part which connected the plunger to the jaw linkages. Any slop in this part would be magnified through the rest of the mechanical system. 3D printers aren’t perfect, and there was some slop — enough that the parts would pinch and bind up while moving.

[Alex] already has a revised design in mind. This is very much a work in progress. That’s the beauty of well documented projects on Hackaday.io — you get to see what works, as well as all the trials and tribulations it took to get to a final working project. Keep at it [Alex], you’re almost there!

That’s it for this week’s Hacklet, As always, see you next week. Same hack time, same hack channel, bringing you the best of Hackaday.io!

Soft Robot With Microfluidic Logic Circuit

September 7, 2016 by Steven Dufresne 17 Comments

Perhaps our future overlords won’t be made up of electrical circuits after all but will instead be soft-bodied like ourselves. However, their design will have its origins in electrical analogues, as with the Octobot.

The Octobot is the brainchild a team of Harvard University researchers who recently published an article about it in Nature. Its body is modeled on the octopus and is composed of all soft body parts that were made using a combination of 3D printing, molding and soft lithography. Two sets of arms on either side of the Octobot move, taking turns under the control of a soft oscillator circuit. You can see it in action in the video below.

Continue reading “Soft Robot With Microfluidic Logic Circuit” →