Let's Make Robots! | RobotShop

Wheelchair Robot for rough terrain farming telepresence joystick question

I am new to the forum and want to first thank everyone for being part of this vibrant robotics community.

I am working on a telepresence robot for long distance farm and greenhouse consultation.  I myself have lived off grid for several years and have experience in farming and creating local food systems.  I however live 350 miles away from my friends new farm so it is difficult to be there to offer hands on assistance.  Enter CESAR, my new robot project.  It would be a great help if I am able to remotely "walk" the farm and see what is going on with the plants so that I am able to assist with changes and plants requirements in real time.  I have worked with UAV and UGV's for some time but I have only recently started working with them over networks and through the cloud.  I have years of experience with that size of power system and the autopilots and programming needs.  However, the telepresence ground robot with a lead acid power system are a bit different and tapping into the proprietary controller seems to pose a challenge.

That is my first question in this project.  I am using a Jazzy Jet 2 as the base for this project.  I was able to find one on craigslist for relatively inexpensive and in pristine condition with great batteries,  I am trying to attach pictures on this post.  I have read several  posts on here and on other sites where people would just tap into the joystick control with an arduino.  Every post I have found whether it is a Jazzy or another model have either a 4-5 wire setup out of the joystick.  I have six?  I put a multimeter to test voltages and it is of course powered to 5 volts.  There is a power and ground wires.  The four signal wires left are centering the joystick at close to 2.5 volts.  There seems to be a hundreths to tenths difference between the four wires.  Two signal alternate volt changes in forward and reverse.  What i mean is one wire will show increasing voltage in forward while the other wire signals decreasing voltage from 2.5 volts.  Of course they are opposite in reverse. The other wires remain static at close to 2.5 volts.  They alternate currents when you move the stick left and right.  Has anyone had to tap into a controller that has had this wire configuration?  Thinking about it even while writing this post makes me think that if I signal two pins with analog voltage signal with proportianally opposite signals that I could make this work.

The second part of this question then becomes.  Should I just work with the controller, knowing that it may take a bit to code to fix? Or should I just rip out the joystick, the brakes and the wiring and install a sabertooth and work with fresh motor control and control software?  I look forward to your responses, ideas and experience.

Comment viewing options

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


I'm also working on a robot powered with wheelchair motors, but I basically am using just the motors powered with an H-Bridge (yes, one of the higher end Sabretooths will work, as will many of the Pololu controllers).

However, please be aware of just how powerful a wheelchai motor powered robot can be. When a person is sitting in it, they are much more aware of their surroundings than you will be with only a camera to guide you. I would suggest having some sensors to turn off the motor power in case of coming too close to an object from any angle. You might have an override, but it's useful to have sensors to prevent running into people, small animals, and off cliffs.

A trivial thing: most wheelchair motors (and I'm including the Jet motors also) have a brake in the back of each motor. This is kept powered to keep the wheelchair moving and is automatically engaged when the joystick is dropped. Taking it out is trivial but for your purposes you might want to leave it in.

The joystick protocol isn't just a normal one for PC joysticks, but can be looked up. Since I'm developing an autonounous balancing robot, this could actually cause me much more trouble than it's worth, so I'm replacing them with encoders. However, if you're using the wheelchair base it might be useful to enable your bot to be stopped very quickly under any number of circumstances since your bot would be stable enough for this.

There are pros and cons to using the existing electronics. on the whole I'd use the existing electronics. I'd remove the seat and stick a head on a pole onto the existing seat support.


1. It's less expensive because everything is already there, including the charging circuitry, 2. You can use the built in safety features including the brakes.


1. You will need to decipher the joystick. 2. You'll have to add your electronic over an existing system.

Good luck and be careful.

that there are some joysticks that do away with the typical pot and make use of hall effect sensors. Your joystick may be one of those set ups.

While reading a forum post, I saw someone mention just adding servos to move the physical joystick.




Farther down the road, mtriplett is happy to point out that a smart phone or tablet has a fair bit of computing power as well as some integrated hardware that could be found useful, if you can get the data to the outside world. If you put a tablet on an arm with a spotlight for those times when the tablet is blocking the sun, you could "get right down" on the plants.

Wheelchair controllers use a specific protocol that is designed for safety.

It is described here http://trace.wisc.edu/docs/set/set.htm

Though since the wheelchair is to be operated as a telepresence robot, which is essentially a large RC vehicle, it might be better to do one of a couple of things:

1. Take the joystick mechanism and use it on the operator's side, if it's in good condition. 2. Sell it on eBay if he doesn't want to use it. 3. Keep it on the wheelchair so somebody at the robot's position can guide it if necessary.

And yes, several people here have used a smartphone or tablet to control robots. One can even use the tablet's screen to put a live image of the operator's face on the screen to help drive home the impression that one is talking to a live person rather than a robot. Also smartphones have cameras, bluetooth, accelerometers, audio, which are sensors you might have to buy otherwise. And Androids are fairly easy to program.

Me, I'd do it with a Raspbery Pi or similar single board computer (Bannana Pi, Beagle Bone Black, Intel Edison) and some small displays. I'm just more comfortable doing this way.

I like the idea of a arm, and there are some interesting ways of controlling it.

For controlling the head, I'm wondering if two cameras on the robot's head linked to an Occulus Rift, which with the proper interfacing should be able to control the position of the bot's head... It would help the operator have the feeling of actually being there.

I know very little about agriculture, but I'm wondering if it might be useful to be able to switch to an IR camera to see the heat...

Thank you for posts.  I really thought that I would like to get into the existing controller but I think in the long run it is going to be better to just go ahead and use the sabertooth to start with.  I have a 2x60 that just arrived today.  I think that I will have greater freedom and flexibility if I just start from scratch instead of trying to decipher the Jazzy controller.  I am still working on a few details for the onboard computer.  I like rasberry pi.  I have been wanting to try out an odroid xu3 or an edison as well.  I will keep working that out.