I had some questions regarding how much computing power would be required to make a small scale 2 legged humanoid robot, that could go running (or 4 legged like those Boston Dynamics robots)...what type of programming (Picaxe basic, Arduino, C?) and board would be best to use? There would be so many servos, a lot of real-time balancing and vision data to process.
Could PICAXE Basic be used with an appropriate board? Or would the processing power be so large that a whole desktop computer would have to be attached to the robot? PICAXE Basic seems like such a straightforward language that it would be very manageable to make some type of AI selft balancing robot. But i've only scratched the surface, and i'm basing understanding on seeing how easy it seems to move servos by writing such little code. (got this from the start here right on this site, what a great tutorial!!)
Also, here's at least 3 new boards coming out that are not using PICAXE basic though:
Would these new boards be able of handle the calculations?
You know how when calculating current, resistance, voltage, we can use that ohm's laws equation? Is there a way to calculate how much processing power one would need for certain robot functions?
I think I'd like to start my robot project with the basic hardware that would be required to process all of the real-time info needed for a self-balancing, walking and running robot. I'll just start with the robot following lines etc, and as I learn more programming, i can just build, trouble-shoot, and integrate new code into what I have, instead of finding out that my setup is limited and then having to buy new hardware and learn a new programming language etc. Its going to take all of my brainpower just to wrap my head around simple programming, and I just want to start on the right path. I'm ready to give it a go and see what type of robot i end up with in 3 or 4 years.
Honda spend decades and millions to get a biped to run so I think you set your target a bit too high to say the least, but don’t let that stop you from trying.
But Honda didn’t have access to the LMR expert database when it started its project. Its just like computer animate films…there was a short film a while back…and it cost 4.something million to make. And nowdays, that type of film can be made in our homes for practically nothing. All that would be needed is for animators to draw frames for a section, and then they could add them up and there’s the film right there.
In this case, each robot maker would have their own robot, but their robot would be using a worldwide and ever-increasing database of motion-correcting knowledge. Instead of one person logging research data over 20 years, there could be 20 people logging data in 1 year, which would be amazing!
I’m currently in my first year (as a mature student) in Industrial Design, because I was tired of working in call centers and I wanted a career change. But its not floating by boat, i just don’t care about designing hair dryers and lamps. Somehow over the past few weeks something has kindled an old interest in electric motors that I had when I was a kid, and now I’m having trouble getting my homework done as i’m spending so much time online looking at boards, chips, robots, and the like.
I think I will not drop out of my program, but finish it and perhaps try to design some gadgets for the micro-robot world. But in the mean time, its time to go to extremes. Its time to show the big corps who the bosses are, its time that artificial riders entered motorsports and win until they are banned. At which point a new class division will have to be opened up, unless the sports authorities allow human and artificial riders to participate in the same sports. Come on I know you want to switch on the TV and watch a Robot tearing up a dirt bike race or snowboarding and doing inhumanly possible tricks!
It’s true that shared knowledge is more beneficial than inventing the wheel over and over again which companies and universities worldwide do, because a robot can have militairy benefits. Sure they share to a certain extent, but a lot remains secret. And for what? Robots that will fight (or aid in) wars eventually? A load of bull****. War needs casualties, one way or another!
Back on topic: It’s true that “amateur” robot builders together can build quite sophisticated robots, but they’ll lack the goodies companies have. Laser range finders, high speed cameras, etc. Even if you’re able to buy state of the art parts there’s always the limit of funds. I don’t have 100k or more to spend, I don’t even have €1000 to spend in fact, a pity really.
I just finished looking at outrageously priced laser range finders, right before reading your reply, which sums everything up. There are the necessary “goodies” out there to make a super humanoid robot, but its just way too costly right now. argh. I’m going to have to think of some other interesting plan.
In the mean time, is there a way to use a computer to directly control the things that an arduino board would have controlled, like LED’s and servos? I know that usually a computer is connected to an arduino while code is being tested out, to move servos etc, and then the sketch is sent to the arduino board so that it can move those servo’s by itself.
Is there a way to take the arduino out of the loop and have the computer and servos only? If so, I guess the servo’s would still need to plug into something. What type of board would that be called? Thanks!
Even I studied Electronic Engineering and have several years of experience robotics is a sum of different types of skills: from electronics to computing also mechanics and even cost management ;-), so it’s better to start small, knowing and testing these elements and then growing up your experiments adding more servos, sensors and power. You have already the most important things: your will power and this formidable team that is Let’s Make Robots
Good luck and I would be pleased to follow your developments.
You’ll need some kind of controller to be able to control servos with your computer. I bought my Arduino Mega 2560 clone for less than £10 on Ebay including USB cable and shipping. Besides having a whole lot of GPIOs and analog inputs, it has 15 PWM pins for servos, motors etc. I don’t think you can get a controller much cheaper.
If you don’t feel like learning the Arduino dialect, just copy the code you want from the internet alter it to your needs and upload it to the Arduino. Controlling the Arduino in your favorite programming language is easy, just send your commands over the USB port.
I couldn’t find a deal that low on ebay, but i found one on fasttech.com for $14.72 incl USB, with free shipping. So I think i’ll pick that up and start experimenting, and basically get this journey underway.
hey exwhyzed, have you achieved anything so far? I have made some research in that direction and I agree with you that it is possible. Maybe we should discuss this?
The challenge with running is actually not the calculations, which can be even easier than for normal walking, but the actuators. You need to react really fast and with really large force, and I’m afraid the electric servos available on the market are not up to that challenge.
There is a recent new development, a robot called Cheetah (not the Cheetah Cub) from, which can properly run and even jump, thanks to custom pancake motors they made for it. See this for details: http://news.mit.edu/2013/mit-cheetah-robot-0308
Thanks for the feedback. I have checked the Cheetah and it is indeed good. I thought about the motors and their limitation as well. But the problem I see with all the current bipeds it is just wrong approach of mechanical motion but not balancing motion. You should start with a self balancing stick, 2 segment stick, then 2 of 2 segment sticks etc… and all this should be self-balanced as human beings do usually. There can be limitations in computing power (but I don’t think there is something heavy to compute, instead just read the IMU sensor fast enough and of course being able to actuate motors.
And then, again, who said the robot should run right away? can’t we just start with walking? but real self-balancing walking would be enough and for that the materials (motors, sensors) are here. It is just noone have yet build a good biped (except darpa maybe, but that thing is heavy as hell and probably got so much of legacy code that can be simplified and if “written properly from scratch” can be achieved similar results.
As for walking before running, those are two completely different problems, and running is often actually the easier one. It’s a little bit like riding on a bike – riding fast is much easier than doing it slowly.
Thank you very much for the links. This I did not know and see it is indeed cool and not only darpa does this. But still, I want a balancing biped. Will start doing it and will see how it goes… Most likely a project for couple of years… If someone helps it should be faster…
Walk normally around the room. Now, using the same motions, slow it way down. It is impossible not to fall.
Animals, including us use various gaits. And the gait used depends on the speed. Momentum is used to carry through the step.
Even walking requires a certain minimum speed, which will be very difficult to a chieve with the equipment affordably at hand.
None of the hobby bipeds you see do a normal walk. They do something closer to a shuffle. In human slow motion it is the foot that keeps us upright, that is why all those hobby humanoid bipeds have such large feet.
Take a look at Asimo and you will see that Asimo walks with knees bent. The problem with bipedalism is not in the balancing, it is in the geometry and also speed. Self balancing by itself will not make a humanoid walk.
It is my opinion that you need to get the physical architecture right, and you need enough power to drive the gaits desired. Processing power alone will not get you there.
Walk normally around the room. Now, using the same motions, slow it way down. It is impossible not to fall. Animals, including us use various gaits. And the gait used depends on the speed. Momentum is used to carry through the step.
So what do you want to say with this? I worked as a professional animator for more than 3 years (and not counting time I spend doing it at home). I have created many walking cycles (and many gaits) and of course there are many types of them. And you can walk slowly and not falling. In any case all walking motions have one intention - keep a structure (human body) in balance, so we are not far from a typical pendulum task, it is just it requires more elements to take into consideration.
Even walking requires a certain minimum speed, which will be very difficult to a chieve with the equipment affordably at hand.
Again, it would be more helpful if you can just suggest motors instead of saying how impossible it is to do…
None of the hobby bipeds you see do a normal walk. They do something closer to a shuffle. In human slow motion it is the foot that keeps us upright, that is why all those hobby humanoid bipeds have such large feet.
That’s exactly what I am talking about and I am saying that it is complete shame that so many smart people can not do a stupid walking cycle based on balance. And therefore I have to start to do it by myself (inviting others to join) as I am sure it can be done (at least to a physical impediment where it can not be done because of lack of available motors or computing speed, but again, we will know that impediment and will start to think how to solve it).
Will start doing it and will see how it goes… Most likely a project for couple of years… 