Building a Robot for work - Require help with controlling it

Projects General Guidance
1

Hi all

Been a long time fan with the idea of actually building a robot, just never had the money or time to do so.

Now I've been employed by a Air-conditioning cleaning company I chance to

We currently use a robot to do the job in ducts that we cannot physically enter, the problem with this robot is the fact that:

its rather limited in climbing
the cameras failed and are not replaceable
controlled by a handheld joystick
it costs at least $20,000

I got talking to the boss and he has said that if i can build a robot that is:

controlled by laptop via USB
moves by tank tracks
has front and back cameras with dimmable lights
and dosent cost more than $1,000

he would pay me to do so

so far i have worked out how the motors will drive the tracks, how cameras will be mounted and viewed and how the case will be done, however i dont know how to control it

so really im just after help on what i should buy and what program i can use to drive the thing and dim the lights

must be via a USB cable

please and thank you for any help guys

ps: here is a link of the current robot in action

2

How long will the cable be? Might want to change the communication to rs485 or something made for long wire runs and maybe noisy environments. There are chip and shields to make 485 relatively painless. A Uno on the control end with joysticks and another on the robot to receive and control motors and lights. Are the ducts round or square or both?

3

anywhere between 5-20 meters

ducts are a bit of both

was actually just thinking maybe i should use Cat5, power will supplied by a wall adapter

just found the phidgets SBC3 [is there something like that from arduino], something like that would do just fine if i can work out how the inputs and outputs work and get some form of code going for it

4

Got a link to "phidgets SBC3" ?

5
6

  1. Go with ethernet - use a cable with an integrated pulling line (in case you need to pull on the comms line to help dislodge or otherwise remove the robot).

  2. You didn't mention if you planned on re-using the existing robot chassis (???).

  3. If you are going to use a "new" robot chassis - you might try out this one (Pololu - Dagu Rover 5 Tracked Chassis with Encoders) - it probably isn't too big.

  4. You'll need to add motor drivers - this one (Pololu Qik 2s12v10 Dual Serial Motor Controller) should be ok for that chassis. They also sell others; look around.

  5. An alternative chassis (but maybe not as sturdy for the task?) - Pololu - Tamiya 70170 Remote Control Construction Set (crawler type)

Note that while I am showing Pololu products - that doesn't mean there aren't other alternatives, nor am I endorsing them specifically - but they do make some great products!

  1. Another alternative (though more expensive) chassis might be to use an Traxxas E-Maxx or similar R/C vehicle - they are designed to be tough, and they are easily controlled using the Servo library. Note that 1/10 scale vehicles may be too large - you might want to go for something smaller (which will also typically cost less). Just stay with actual hobby R/C devices, and not toy ones - they are easy to control, though more expensive.

  2. If you must use a toy R/C car chassis - check out this link: http://forum.arduino.cc/index.php/topic,86883.0.html

Now - basically what you are going to do is directly control the robot chassis using an Arduino; it will likely also have an ethernet shield or similar interfaced to it. Once you have control via the interface, the Arduino can send commands to the motors, servos, relays for lights, etc - whatever you need to control.

By using an ethernet shield (or similar) - this opens up an important need - the camera. If you plug all of the ethernet stuff into a small 5-port ethernet switch (NOT A HUB) - you'll have on the switch the Arduino, the tether back to the PC, and the camera. Oh - the camera...

Use a cheap pan/tilt IP camera - like this one:

http://www.amazon.com/EasyN-Wireless-Camera-2-Audio-Nightvision/dp/B005O3YS38

Most are super stupid simple to control, usually via a simple web interface (you send commands to a certain URL and it does it) - getting data (video) back is also very simple - it is usually an MJPEG stream that you just point to with another URL. Note also that many/most have IR "night-vision" capability - perfect for dark duct-work!

So - all of this is plugged into the switch, then the cable back to the PC. Also note that your robot must have enough power to drag that cable around - keep that in mind. You might (honestly) experiment around with using a hobby R/C chassis with a 2.4 GHz radio - and see if you can communicate using the radio inside the ductwork. Attach a thin (but strong) spool of monofilament to the car, though - in case you need to pull on it to rescue or unstick it.

On the PC end - you would need to create a simple command and control interface - the easiest (well, in my mind) would be to set up a LAMP/WAMP/MAMP system - point the web browser at the local web server instance, and it would send commands using TCP/IP to the robot, and receive data (video) back in a custom web browser interface. Open the IP up, and you could control it with your phone, too - if wanted. If you use WiFi (instead of tethered) - you could possibly put the web server on the robot instead - and control it that way (and have the WiFi send the video back).

All of this is quite do-able; you are a lucky guy to get this opportunity! You might want to clarify with your boss on who will own the IP and construction rights to the robot - because if you can do this kind of a robot for under $1000.00 (and I think you can do it for much, much less) - you could sell this to not only your boss, but to other companies and such as well that need to do inspections (ductwork, pipes, attics, crawlspaces, etc). If your employer was willing to drop $20k on such a robot - others are as well. And they will be all over you if you can offer the same capabilities for a fraction of the cost. But if you don't get that clarified (and in signed and notarized writing) before the fact - you'll find you did a ton of work on something that could have let you retire - and instead that work goes to line the pockets of the company's owner and/or your boss.

Think about it...and good luck with the project!

:smiley:

7

he would pay me to do so

As previously mentioned, he will own the rights to what you develop unless you have a written agreement otherwise.

8

for the chassis its most likley going to be just a alluminuim hobby box with some fans and filters [duct i will be going in has a build up of at least 20 years of dust and mould so will be sealed]

will be controlled just by an ethernet cable, tied together with a power supply for it, and twitsted on some rope and covered in tape of some kind so its protected and pullable

cameras will just be some USB web cams

will be driven tank track style as it allows for small turning circle

if i can get it setup to be controlled by a custom controller [just a box with a single joystick and potentiometers to control diming of the lights] and the webcams viewed on the laptop that would be perfect

i shouldnt have bludged during ITS at school, cant code for me life lol

9

zoomkat:
As previously mentioned, he will own the rights to what you develop unless you have a written agreement otherwise.

I do not believe this is true.
if you make something while under employment, using both finances and resources from the company, then you are are making it as part of your employment. but, the actual person doing so has the right to have their name on the patent, even if the patent is owned by the company.
however, if you make a thing at home, on your spare time, for us in the industry in which you are employed, you have the full right of ownership.
if the boss said, here is $1,000 show me what you can make, then the boss has hired you for the project and probably owns the rights. if the boss said, here is a nickle, it is the same in the eyes of the court as if he gave you $5,000 as it was a legal contract, with financial backing.
the sooner you get the agreement, in writing, the better. if you have a conversation with your boss. you can write an e-mail and review the points of the conversation.
I would not use ANY free legal help. this site or others. if you think you can turn this into a business.
and your company lawyer works for the company, not you. they cannot work for you, conflict of interest. anything you tell them, they will use against you if they can.

10

google sewer camera pan tilt

you can get lots of stuff, high quality for $1,000

11

BrownDog186:
for the chassis its most likley going to be just a alluminuim hobby box with some fans and filters [duct i will be going in has a build up of at least 20 years of dust and mould so will be sealed]

will be controlled just by an ethernet cable, tied together with a power supply for it, and twitsted on some rope and covered in tape of some kind so its protected and pullable

Based on this description (and little else) - I can't imagine that the robot will have enough traction or power to pull along any great length of such a cable and safety line. Also note that ethernet cable (that is, the usual stuff from a box) is made of solid core wires; stranded core cable does exist, but it isn't cheap.

This is why I postulated a wireless approach - it would allow the robot to be smaller, and lighter weight, yet still gain you plenty of functionality (assuming that it worked inside the ductwork, of course).

BrownDog186:
cameras will just be some USB web cams

How do you intend to get the USB signal back? You can't just wire it "direct" - USB is only good for a few meters of cable; you'll need boosters/amplifiers, etc - and you'll need a hub on the robot end to connect it all up - unless you plan to put something on that end (like a small embedded computer like the Raspberry Pi) that can talk USB to multiple devices and send it back along one way. Note, though, that the Pi's USB isn't actually done with multiple ports (IIRC) - but rather is like a "hub" - so speeds and other things when using multiple devices will be affected.

This can all be avoided if you go with ethernet and a switch.

BrownDog186:
will be driven tank track style as it allows for small turning circle

That's a good plan for such a robot. But note that you could also get away with four-wheel differential skid-steer, which might work just as well in the task, and be easier to source parts for (mainly the treads - sometimes those are easily available, other times they aren't).

BrownDog186:
if i can get it setup to be controlled by a custom controller [just a box with a single joystick and potentiometers to control diming of the lights] and the webcams viewed on the laptop that would be perfect

Perfect for the beta, maybe - but if I were going to develop this as a product for sale to others (let's be honest, I bet that is what your boss is thinking of doing) - the controller (and recording, etc) - would all be done via a smartphone.

Which - again - wireless ethernet (802.11) would be perfect for.

The more I talk about this, the more I think that I should get off my duff and just build one and market it myself to local inspection businesses and individuals; it's something I've thought about for a long while, and I have all the skills to do it - and heck, I probably have all the parts laying around in my shop to boot. I'm just a bum, I guess...lol.

12

I do not believe this is true.

It will be what ever the courts decide under under existing laws and contractual agreements.

13

The below lynxmotion robot platforms might be of interest. They also have the track material and sprockets to fit for custom building.