Some of you may remember Brainbot, a project directed and funded by the Brain Engineering Laboratory and Neukom Institute at Dartmouth College, designed in collaboration with Jon Hylands and HUV Inc. The purpose of which is to enable the study and practice of brain engineering.
Well, Brainbot is getting a brain transplant into a 1/5 scale top of the line rock crawler chassis, for the purpose of creating a navigation variant. Onboard smarts? A Core 2 Quad, 2.83 GHz, 2 GB RAM, and a 32 GB Solid State HDD, 802.11n wifi, GlobalSat USB GPS, Prosilica gigabit ethernet camera (640 x 480, 60 fps) which plugs into Roborealm, and a Hokuyo UTM-30LX scanner laser rangefinder.
I went out and bought a 1/5 scale RC rock crawler chassis. Just for scale, those tires are 8″ in diameter, and the whole thing is 30″ long. We chose this chassis because it would have no issues going long distances at speed, and it would be able to handle curbs and grass and such with no issues. Being a rock crawler, it also goes slow enough that the sensors can keep up…
Right now, it has the quad core mini-itx mounted, and is being driven around with my Logitech wireless joystick. The motor driver (a Sabertooth 2×25) is controlled from one of my Bioloid I/O boards, which is connected to the bus. A program (written in Squeak Smalltalk) takes the joystick input, and sends speed and steering commands over the bus to the I/O board and the steering servos (AX-12’s).
Dr. Dennis Hong, who is in charge of the RoMeLa program at Virginia Tech recently made a presentation at TEDx showcasing some of his department’s very advanced, very cool robotics projects! I couldn’t help but notice watching through this rather lengthy video (it’s worth watching all of it, trust me) that a good number of Dynamixels from Robotis made an appearance. Check it out!
Our good friend Jon Hylands, of HUV Robotics (maker of all those awesome 3rd party Bioloid accessories), has recently started a new 4-legged Bioloid based walker named Roz. He is currently using an arbotiX Robocontroller, but has plans to eventually migrate over to a custom Gumstix-Overo platform. He is utilizing the upcoming NUKE (Nearly Universal Kinematics Engine) on the arbotiX, which will be available in public beta in the near future, and is really coming along in his walking gait progress. This has to be one of the fastest AX-12+ servo based walkers I have seen yet, check it out!
SMALdog, or Sorta Maybe Almost LittleDog, is Mike Ferguson’s (creator of the Arbotix Robocontroller) latest addition to his AX-12+ based quadrapod robot lineup.
Rather than utilizing the more common ‘lizard-style’ leg configuration, SMALdog has a more ‘mammal-style’ leg configuration that provides some interesting walking gait variations. The biggest difference is even with very little gait programming done so far, it’s already faster than the award winning Issydunnyet quadrapod. There is still a lot of work to be done on refining the walking gait, and the turning gait is barely functional, but this platform certainly shows promise. Check out Mike’s thread for more information and updates!
I won’t even pretend that I understand all that is going on in the walking gait here, but this is darn near human-like in every aspect. And being able to walk like that is one thing, being able to recover from a sideways push is an entirely different accomplishment all together. Absolutely brilliant.
Looks like iRobot is determined to bring Judgment Day upon us. First with the military bots, and now they’re trying to build a liquid-metal-shape-shifting robot that will surely ask us if we are John or Sarah Connor. Okay, not quite, but it looks like they’re headed in that direction.
This week at IROS 09 (Intelligent Robots and Systems), iRobot and the University of Chicago unveiled a soft, blobby robot that looks something like an inflating marshmallow.
The new robot, called chembot, changes the shape of its stretchy polymer skin using a technique called “jamming skin enabled locomotion”. This means that different sections of the robot inflate or deflate separately; controlling this inflation and deflation enables the robot to move. DARPA, which is funding the project, hopes to use the robot to squeeze into small holes or under doors, which I’m guessing would be used for sophisticated surveillance.
The complexity shown in LittleDog’s adaptive gait and terrain navigation seriously makes my head hurt. If you consider yourself a geek you are of course familiar with BigDog; the 4 legged robot that robot dreams are made of. LittleDog is similar in concept but at a much smaller scale, and MIT has been using it as a research platform for a number of years. Evan Eckerman from Botjunkie sums it up quite well:
Researchers at places like MIT have been usingBoston Dynamics‘ LittleDog robot for years now as a testbed to teach legged robots to learn how to traverse variable terrain on their own. This video shows some highlights of a “dynamic double-support gait,” which means (as near as I can tell) that LittleDog is supporting itself, at times, on only two of its four legs. This is a substantially more efficient way of negotiating terrain thanwe first saw two years ago. LittleDog also demonstrates some markedly biological ways of negotiating obstacles (with the possible exception of the belly flop on the Jersey barrier)… I especially liked how it pranced in place slightly before tackling each stair. All this stuff is obviously a lot of work for a little bot, since poor LittleDog completely collapses at the end of every test.
LittleDog, remember, is teachingitselfthe most efficient way to negotiate these surfaces. Overhead cameras examine the terrain and plan out LittleDog’s route by computing a ‘cost’ for each step, which takes into account the distance moved towards the goal as well as the potential for a fall. After a lot of trial and error, LittleDog figures out how to best compromise between progress and stability, and the lessons it learns could be propagated up toother, larger quadruped robots.
This video is from Phase 2 ofDARPA’s Learning Locomotion program… MIT’s LittleDog team was awarded funding for Phase 3 of this program back in 2008, so we’ll keep you updated.
Who doesn’t like free stuff? XMOS is offering some development kits of their new event driven parallel processors to qualified parties and individuals. These controllers looks like they pack a punch in terms of features and processing power and would make an awesome robotics controller! If you’re interested, check out the forum post by XMOS and answer the requested questions. Definitely shows some promise, check out the forums and the videos below for more info.
XMOS event driven processors allow you to execute code in parallel which could open up a number of possibilities to evolve your robots “mind” to the next level!
I have some development kits to give away potentially, but before I give away all the details, I just wanted to see if there was any interest.
People have made some cool stuff with our tech before (See videos).
Some details of the processor itself which will be on the dev kit (to be released soon):
- Single core device (Although we do have quad core versions – ask me)
- 400 MIPS per core.
- 8 Threads per core.
- 64Kb RAM
- 8KBytes OTP memory for applications, boot code or security keys, with security mode
- 64 user I/O pins
- Support for high performance DSP (32 x 32 → 64bit MAC) and cryptographic functions
- Time aware ports provide up to 10ns timing resolution
- Designs implemented using a software-based design flow (can program in C or XC – very similar to C but with support for extras such as parallel execution)
- Scalable – can connect many kits together for crazy amounts of processing power using Xlinks.
Sure, the researchers may state they are intended for assistance in disasters… but these wifi network creating robots could just as easily be used to spread the reach of the dreaded Skynet! I have my doubts, but Engadget seems pretty sure that these robots have good intentions. Time shall tell!
Researchers at Germany’s Ilmenau University of Technology are developing flying quadcopter robots that can be used to form a self-assembling ad-hoc wireless network in the event of disaster. Built with off-the-shelf parts (including VIA’s Pico-ITX hardware and a GPS unit) the robots are designed to provide both mobile phone and WiFi access — and they can do it far more quickly than a technician on the ground might be able to. The device comes in a kit for €300 (about $380), which includes all but the battery — the batteries currently run around €1,000 (over $1200) and only offer up 20 minutes of flight time. Once the device has found a perch, however, it can operate for “several hours.”
Our good friend Jon Hylands (who is an amazing Roboticist, definitely check out his work on the Bioloid system) is working on a few new projects and shot a quick video which is pretty impressive to anyone who has ever built a robotics arm. What you see in the video below is an RX-64 Robotis servo lifting a full coke can at a 30 cm arm length. While that seems trivial to we humans, packing that much torque into such a tiny servo is a great demonstration of how much servo technology has advanced in the last few years. The servo isn’t even being strained during these tests. You might note a bit of jerky movement, that is because the servo is being manually fed positional instructions via a slider bar. Most advanced robots use some form of interpolation which smooths movements considerably.
We’ll also be revealing our big internal project which uses 14 of the RX-64 servos in the near future… =)
