Let's Make Robots!

Stepping up/Moving Forward...

First, via adafruit (who has apparently bought an Italian URL so that they can shrink urls using "adafru.it") I read a great think piece today by Leila Takayama at MIT Technology Review about what it means for a robot to be human friendly.  A few quotes:

"A more important objective, to my mind, is making robots more human-friendly in their form, behavior, and function. By this I mean that robots should be appealing and approachable. They should behave in ways that are easy for humans to interpret, and they should perform functions that meet human needs."

"Finally, these human-friendly robots must meet real human needs, not only the needs of their inventors. Fetch-a-beer and fold-a-towel demos are nice scientific steps toward building more general robotic capabilities. But what we need now is for human-centered-design researchers and product-minded entrepreneurs to do the dance of the necessary and the possible with the robotics community."

Takayama goes on to lament that most of the Robots deployed to human environments are a temporary fixture so far, and that the real challenges aren't so much in making robots "social" as they are in making them compatible to humans and human environments.  That may mean designs that don't fit in with our preconceptions about what a robot should look like or do.

If I can dip into the world of fiction for a moment, specifically poorly adapted screen sci-fi, the Will Smith version of "I, Robot" (which had almost nothing in common with the Asimov book of course) demonstrated mostly by feel the problem here.  The Robots that were introduced into the home environment seemed almost to present more of a problem than a solution to anything just by their presence, and that was before the main hell broke lose.  They took up as much space as another human would, they did piddly little tasks that really didn't save much time, they annoyed their owners.  I don't know if it was the film-maker's intention to present them that way, but he/she certainly stumbled on a problem that may have to be faced:the problem with using robots to replace humans in day to day tasks is the replacement of humans in day to day tasks may not be as fun as we think.  Okay, way off on a tangent...

For robots to really take hold in the world in a useful way, they really have to be doing something necessary and useful.  They also shouldn't be terribly obtrusive.  You probably need to be able to shove it in the closet most of the time.  Chances are there won't be one all-purpose "maid" or "butler" robot-your refrigerator isn't your laundry unit, after all. Chances are the transformative robotic product won't be a half-million dollar PR2 coming into your home, nor will it be something that evolves out of the Roomba.  But, slowly it will become obvious that every household can experience benefits from two or three mobile specialized computers with sensors and actuators of some kind.

Or maybe not slowly.

A while ago in a thread that was, if I remember right, dedicated to the BeagleBone Black, someone (bdk6, merser or mogul maybe?) said we needed to "step it up" when it came to what we were doing with robots, and that struck me as being very true.  I've always been most interested in building robots that serve a useful purpose.  It's harder to do than it sounds, but not beyond what most of us who have been around LMR for longer than it takes to figure out why their line-follower isn't working are capable of.

We are a very talented group, and together we push evolution.  I'd like to see us make that a little more concrete of an assertion is what I'm saying here, I guess. Baby steps are just that, but we've already taken most of them-how many variations on sumo-bots and hexapods can we build and where do they get us?  We know how to manipulate the technology, it's time to start making it work for us as more than just amusement.

That's not to say I know what the next step is.  Certainly my household chores are limited-"lawn bots" are commercially available (though more expensive than would be worth for my lawn) and a laundry folding robot is years out in terms of price and utility.  I can get my own beer (and let's face it, if you're an American you probably shouldn't be grousing about getting off your butt and burning the calories!)

So bounce it around in your head a little:what, in your home, would you like to automate?  Where would a robot be useful and not just a toy?  Do most of us even really have it so rough that it would be worth the space we'd give up for a minor convenience?  Or will the home robots of the future even be mobile?  Will they be arms recessed into the walls that come out like an ironing board when needed?

Personally, I'm a little ashamed to say that all I can come up with right now is that I wish I had a machine that could do my jogging and waking up in the morning for me.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Dreamers all of us aren't we?  Imagine the things we can do and what a difference we can make in the world.  I truly see technology as a force for good.

The reason that people make simplistic robots is the software interfaces to hardware are not there.  It is an accomplishment to get a robot that moves around and avoids obstacles.  Behavior and hardware need to have an easy way to be linked as well as to be able to communicate easily between different cpus in a standard way.  The hardware is here, we need to design and then write the software for that brave new world that is almost on us.

Manufacturers need to sell more than just hardware but the software integration.  Dagu is doing a great job with hardware piece but not so much with the integration from what I can tell.  Perhaps there isn't a market yet for this; I am sure Dagu knows their business.  But if robot builders could have complex algorithsm at their finger tips to easily merge with hardware, these building blocks would bring about more and more complex robots.  Pick and choose your hardware, your software, couple them together and imagine what cool things we would have here on LMR!  Oh, and the manufacturer who does it, sells that much more hardware.

Willow Garage with their Robot Operating System are doing some cool stuff.  I plan on digging into it soon but is mostly Linux.

I posted several months ago about writing a cross platform library that could be that bridge or a step forward.  I completed a proof of concept for Arduino recently and posted this as a blog.










You are correct that as I am not a programmer, DAGU products only come with basic Arduino sample code.

Time is a big problem, even if I was a great programmer, by the time I have designed the product, written the instruction manual, designed the artwork for the box, provided tech support for the distributors (which takes more time as the range of products grow) and provide tech support for individual customers here on LMR I don't have time for my family, let alone a week or two to write nice software and tech support that.

Finally there is the fact that I try to create educational products. If I write software so complete that the customer just has to press a few buttons then the product is nothing more than a toy and the customer learns nothing. My software is only intended to be a foundation for the customer to build upon.



I assumed you were a much larger company working with a team of 4-5 engineers, a marketing team, a support team etc the way you are banging out the cool hardware.  The Wild thumper, doodle bot, that mechanical arm plus tech support, packaging etc etc- wow.  That is all you?  I am impressed.  You have done some impressive stuff.  I meant nothing personal, but meant it as a general observation of companies in the robotics sphere.

I see what Willow Garage is doing as a great model for a company.  Here is our software, here is our hardware - this is how you can link to us, here are some standard algorithms for real world problems.  You want to talk to an Arduino or run our toned down ROS on the Arduino, here is our library.  This fosters a collaborative atmosphere where companies can start up based on Willow Garage hardware AND software and go in a direction Willow Garage never thought about or ever wanted to go.  And it also sells more hardware long term.  Innovation builds on innovation and the bar notches ever higher.

I write software for a living and when I see a nail, I tend to want to hit it with software.  I acknowledge my bias.  But from what I can tell, most companies in this robotics sphere when they see a nail, they tend to hit it with hardware.  CPU cycles in 32 bit micro controllers go unused on robots because the software isn't there to do common things.  Every PID loop is written by hand and by scratch each and every time.  How many times has someone asked about a PID loop they are trying to write on this forum?  Every time someone wants to have that robot be able to mow the lawn, they have to write their wavefront and dead reckoning algorithm from scratch.  Every time someone does a hexapod, they spend huge amounts of time doing Inverse Kinematics algorithm based on a couple of formulas gleaned somewhere on the web.  People make pieces available here and there, but it is a hodge podge and very hit or miss and the quality is sometimes suspect.  My experience has been that the integration of someone else's C code is often more work than actually writing it.  Well designed, object oriented libraries with a comprehensive approach are needed.  Willow Garage ROS is a great start.  Some really tough pieces are now available for Linux such as speech recognition, computer vision (Open CV) etc, and now with Willow Garage, ROS, it really could be the start of yet another revolution where we as roboticists (ok, you all as roboticists - I am so new to this and have so much to learn) have the opportunity to make a positive difference in people's lives through technology. 

If we want a higher bar, if we want to truly step up to the potential to achieve greater things in robotics besides the line followers, photovores, et alia, the software needs to be there.  The hardware is here and at a very reasonable cost.  We need software that is worthy of the hardware existing today.



I did not take it personally but felt I needed to explain why we don't have super fantastic software. We are a small factory and as the only westerner here I do all the English instruction manuals, software etc.

We do have a small team of engineers. One engineer designs the PCB's after I give him the schematics and layout. I could do it myself but I need the time to learn new software. He also writes some software for our Chinese customers.

We have another engineer who take my prototypes and sketchup drawings and draws them in solidworks so that we have files suitable for the factories to make molds and dies. He also advises me on what the local factories can do.

We do have a software engineer but his experience is centered more around android. He works mostly on software for our Chinese customers and is just learning Arduino.

Yes I think Willow Garage is doing cool stuff but how many hobbyist can afford a robot like PR2? My focus is on educating the next generation of robotic engineers. We try to produce low cost chassis's and controllers that are easy to use.


I think you are refering to something I said. It was along the lines of:

"Why get a more powerful, 32bit processor when most people can't even use an 8 bit processor to it's full potential".

Basically, the few robots I have seen using more powerful processors are still just doing basic object avoidance and line following. The only thing they might use the 32bit power for is object recognition with a camera. Even then that can be done with an Arduino and a video experimentor shield.

The reason I did the Tic Tac Toe challenge is because I was tired of seeing obstacle avoiders and line followers. The Doodle Bot art challenge is the same. Just to try and get LMR members to do something different.

Probably what we should be looking at is a robot that can help the elderly or disabled in some way.


That's a great link. I hadn't seen this particular shield before, and it obviously has a lot of power. I was a little disappointed that it uses an NTSC signal instead of a USB Protocol, but on reflection there would probably be no getting around driver problems. If I had a job I'd probably have ordered one by now. I don't know that it could do face recognition or that an UNO could handle object recognition on the level of openCV, but it would be fun to see how far it could be pushed and applied to hobby robotics applications. It seems to be limited to monochrome capture, but still, that would be a challenge worth stepping up to.

Using this sheild with a suitable camera is really just going up a step or 2 from my IR compound eye. It can see over a greater distance and even using simple Arduino code it should be possible to recognize basic shapes.

So no, it won't replace a 32bit ARM processor and camera but for hobbyist and students it would be a good intermediate step.

There are lots of efforts in this arena, especially in Japan, where caring for an aging population yada yada...  That was partly what I had in mind for Yubin Kun.  I wanted something that could pick up my socks and underwear while I recuperated from the surgeries/radiation.  (It didn't turn out to be as useful as I'd hoped, but I was still proud of it.)  Helping the disabled robotically will probably take the form more of cybernetics if we draw the distinction, but I don't know that we should.  I think that your proposal is a noble one, OddBot.  It's tempting* to try to work it into a challenge of some kind.  In fact, I've never examined the challenge archives-has a "help the disabled/elderly/infirm" challenge been issued within memory of the current crop of LRMtians?  Certainly a functional enough unit, designed for that purpose, once proven would be hard to keep from the public at large.  Sick and well alike are equally interested in having machines do the work!



*I'm not committing to anything, here!

I haven't had to care for elderly or disabled people before so I am wondering when one mentions robots to help care for the elderly or disabled, what types of functions would they need to do? It seems to me some functions would require as much sophistication as is not available yet. Like if the robot was capable of picking up socks and putting them where they belong or folding clothes it should also be capable of fetching stuff from a fridge for a lazy beer drinker.

I vote we string Max up by his toes for posting something in a font size of what looks to be about 6. :P