Wednesday, December 29, 2010

Meet your robotic avatar.

Finally got the robot. The battery's bad, but I've ordered a higher capacity replacement. It should permit nearly four hours of use between charges. I didn't bother to assemble the arms because they're non-functional. Perhaps you can think of some devices to install in their place? A machine that sprays out squirrel feed? A 1 watt laser? A power drill?

Still need money for the solar panels and 4G mifi, but the docking station battery could be the same battery pack I used to power Hampture's air pump so as to save funds. This project might take shape faster than anticipated if I can work out ways to reuse stuff.

Sunday, November 14, 2010

The raison d'etre

Still waiting on donations so I can buy components. In the meantime, here's a look at a similar endeavor, the senior project of a group of students at the University of Houston.

There are actually a lot of videos of similar internet controlled robots on Youtube, the original inspiration for the project.

I thought to myself, what if robots like these became a commonplace extension of the internet? What if for every location you can click on for info in Google Earth, you could also click "drive robot" and personally check it out?

The biggest obstacle to this is cost. The components for building custom telepresence robots make it something only hobbyists with a lot of disposable income can consider. The second biggest obstacle is technical expertise. Public access telepresence robots will never become commonplace so long as they have to be custom built by someone at every participating location.

Rather than give up on the idea, having owned a Rovio for some time, and having played with a Spykee I bought for my Nephew last Christmas, I thought "these may be toys, but aren't they good enough? With a few other off the shelf components anyone can afford and put together, couldn't a shop owner, or a hotel manager, or a principal, or anyone interested in having a telepresence robot available for public use set one up?"

So I compiled a list of the necessary capabilities the robot and support station would need to have, and set out looking for products that fit the bill, and that would play nicely together. The result was a design that wouldn't be very pretty and certainly not as efficient as a custom built robot, but one that everyone and their dog is capable of reproducing at home.

It's my hope that once I do this, and show how entertaining/interesting it can be, copycats will spring up. Using the same products or similar, people who ordinarily aren't interested in robotics because of the learning curve or the cost, each making a physical location accessible by telepresence that wasn't before. And who knows? Maybe the next time you want to check out a theater, hotel, restaurant or even nature park before you get there, you'll do it via robot.

Thursday, October 28, 2010

This might seem a little suspect

....But it does work.Since winter's coming, solar panels won't work that well. But an Earth battery might. I'd charge the robot pretty slowly as you need to drive a lot of stakes into the ground before you get a useful current, but if you can deal with 12 hour recharging periods, this would do the trick. It also keeps supplying power for several months before the soil is depleted. Good idea? Yes? No?

Thursday, October 21, 2010

Robot batteries, and Mifi.

So, I've been in email correspondence with a gentleman who sells Spykee batteries. Not standard, stock batteries as shown here:...but rather, similar packs with the same voltage but vastly greater amp hours. As Spykee uses a pack format common to most RC cars, it's possible to swap in higher-end batteries provided they either fit in the same recess, or can be made to fit with minor modifications.

According to this guy, the standard upgrade is from 1.8ah to 2.5ah. That gives you a driving time of around 3 hours. However he's looking to get his hands on some 13ah (yes, that's 13,000mah) batteries that will fit in the same spot with a small amount of case modification. Nothing drastic, just sawing away at part of the battery enclosure. Assuming the operating time doesn't scale perfectly with battery capacity, a conservative estimate of how long you could operate the robot on a single charge using a 13ah battery is 15 hours.

Of course it would also take much longer to fully charge. These are NiCD batteries, which have to be fully discharged before recharging or you get the dreaded "memory effect", but the robots will be in such demand that I don't anticipate that will be a problem. With the stock charger we'd be looking at perhaps 8 hour charging times. Why is this preferable to 3 hour charging times, you ask?

Well, because provided people are courteous and don't use the robot for too long, there'd be a bigger 'buffer'; you wouldn't have to worry about logging on to find the robot's out of juice, simply because it'd be awfully hard for anyone to use up the robot's entire charge before it got dark. And it'd have just enough time to fully recharge overnight. This means less time spent sitting inactive in the charging dock during the day, and more time being driven around.

Coincidentally, it's also possible to buy an extended use battery for the mifi. It ups the usable time from 5 hours to 15 hours, essentially the same as the robot's battery life. Unfortunately I can't work out any way for the mifi to recharge when the robot is docked, or I could mount the mifi on the robot itself and it would effectively have no range. It could stray as far from the dock as it's batteries would take it. This is a tantalizing prospect and I may do it once just to see how far a journey I can take the robot on before it's batteries die, but for everyday use I'll probably keep the mifi on the docking station so that people can't get too far from it.

The area I've chosen has 4G coverage, and as I've heard Spykee is somewhat laggy over 3G, I'll be buying a 4G mifi. The best deal I can find has a $60 monthly fee, which I can't really afford, so while I rejected earlier suggestions that I monetize this project, I may have to require some small form of payment for the robot's use just to keep the connection operational. It will most likely just be a "seriously, please donate" button at the upper righthand corner of the robot's website though.

Wednesday, October 20, 2010

Gibe monies plox.

I don't have much to offer in today's update. I've been unable to buy any of the new parts for Project Earth Rover, as the only donation since the completion of Project Hampture was for $5. (Much appreciated, by the way.) Hate to go all PBS on you but this project's innately more expensive than Hampture was and it's going to be difficult to move forward without funding. Here's a list of everything I expect the project to need:

Spykee robot: $299
Mini GPS tracker: $200
Solar power kit: $289
Mifi mobile hotspot: $242
(Monthly 3G subscription rate: $60)

The good news is I've been contacted by someone who may be willing to donate a Spykee. I've also found a relatively low cost solar power solution that is sufficient for multiple rovers, and will work with the battery pack from Project Hampture.

I'm willing to pay for the shipping and anything else out of pocket that I can afford. I really love this project concept and I am giddy as fuck imagining strangers who follow this blog being able to log into the robot and pilot it around the forest clearing I've selected. But tragically, like everything else in science, it remains a distant dream until funding materializes.

Tuesday, October 19, 2010

Solar Power

Having selected the robot I plan to use, the next step is to find a suitable solar power solution.

What I need is basically a reasonably priced solar panel capable of charging the same power pack that I used for Project Hampture. It's just a regular car battery in a fancy enclosure with an integrated inverter so you can plug a standard 110v power plug into it.

To charge that, I need this solar panel to come with an inverter that has a standard 110v outlet integrated. Do they all? Do only some? Many "all in one" systems don't clearly show the inverter so I can't tell if they have an outlet. Many include what looks like a cigarette lighter plug and a pair of jumper cables, presumably for charging car batteries directly but of course that's not what I need.The rover battery is pretty substantial for it's size. 2 hours of runtime may not seem like much, but for a telepresence robot with always-on wifi, a webcam and powerful motors it's phenomenal. So assuming it's in nearly constant use, pausing to charge for three hours between every two hours of operation. Let's say 6 hours of charging, total, per day (as it won't get much use when it's dark out) I believe the charger supplies 9 volts. Would a 60 watt panel suffice? Or is that overkill? I can get a 15 watt panel that's the perfect size and is much cheaper, I just don't know if it'll keep up with the rover's drain on the main battery. I'm also new to solar and have no idea how to translate solar power into something with a 110v socket I can plug the battery pack into. Any advice would be appreciated.

Monday, October 18, 2010

The mobile platform.

I'll be using Erector's "Spykee Wifi Spy Robot" as the basis for the project. It's marketed as a toy but I've owned it as well as Rovio, and for the money, Spykee is remarkably robust. It has a much longer battery life (2 hours versus 15 minutes) it has far, far lower latency during control (Rovio is cool to look at but nearly uncontrollable over the net) and better for outdoor use (sturdier, has tank treads, more powerful motors). The portion of Spykee that has all of the electronics in it is pictured below: Not pictured; the "head", essentially a webcam with integrated LED flashlight. The rest is the non-functional upper body and arms which I won't bother to assemble as they add nothing and make the robot easier to tip over.

Unlike Rovio, Spykee doesn't rely on being indoors for the navigation system to work. Rovio uses the "Northstar" system, which tracks it's position by IR lasers pointed at the cieling. If it isn't obvious, while I love Rovio's aesthetics and added functionality (the movable head mainly) it's the wrong robot for the job, and a somewhat shoddily built one in general. All things considered, if we're to have multiple robots they need to be affordable and capable. Spykee fits the bill on all accounts. At least, a stripped down "base plus camera" version of him.

I'd also like to add some form of manipulator to each robot. Has to be passive as the system doesn't really permit the integration of complex robot arms or anything like that. Think bulldozer shovels, prods, picks, that sort of thing. Suggestions welcome. Oh, and you may want to download the Spykee control software in preparation. He's currently controllable via your iphone/ipod touch, your PC, your Mac, and your Android phone.

Sunday, October 17, 2010

New project, new blog.

Although I still plan to expand Hampture, I have too many ideas to spend all of my time on a single project. This is an idea I've played around with for years; a telepresence robot that you can control over the internet, stationed in the middle of a forest, using a solar panel/battery pack combo to keep the robot charged as well as the mobile 3G/4G hotspot that provides connectivity.

What if you could explore a deep forest at work? Or between classes? Would you draw what you see? Do battle with squirrels? What if multiple people could explore the woods together, via their own separate robots? Would you form expedition parties? Pester racoons? Challenge bears? The potential here is pretty obvious, I think.

True to my word, I completed Hampture with your generous support. I did the research, assembled the components and the end result was everything we all hoped it would be. If you support this project, rather than simply witnessing the fruits of your contributions, you'll be able to directly interact. Everyone involved will be able to log into one of several robots and go on grand adventures in the great outdoors.

So let's get started, shall we?