finished mowbotWe were only about one month into the 2013 lawn mowing season when we decided that we had had enough.  Mowing the lawn wasn’t fun.  It’s hot, dirty, smelly work that is better left to someone, or something else.  We decided right then and there that we needed to make a robotic lawn mower.  One that could be controlled by remote control, or if we were adventurous enough, one that was fully autonomous.

We had only one requirement for this project.  The robotic lawn mower would need to mow in a pattern similar to how a human would mow – with straight lines.  A lawn mower that operated like a Roomba in a crazy, random pattern was not going to work.  We planned to start with an existing lawn mower, modify it for remote control, and then eventually include an inertial guidance system and microprocessor to allow it to mow on its own.

old mowerThe first lawn mower we had planned to use was an old gas-powered mower.  After considering how this mower would be used, we wanted the ability to turn on and off the blade remotely.  This meant a gas engine wouldn’t work.  We ended up stripping everything off the mower deck, including the gas engine, with the intention of adding a new electric motor and batteries to turn the mower portion into an electric mower.

After researching how much work it would be to add a new electric motor, and how short the runtime might be, we decided instead to use an old Neuton CE 5 mower with a 24 volt DC battery.  We could leave most of the Neuton mower intact since it already had an electric motor, and just wire it up for remote control.

There were two web sites that were very helpful in this construction process.  The first is Robert Smith’s SCRCLM (Solar Powered RC Lawn Mower) project.  Although we did not choose to use as much aluminum, the wiring diagram and parts list from this site proved very valuable.  The second site was a tutorial by John David for and Arduino powered RC lawn mower.  This site was valuable for ideas on how to construct a rollable chassis.  In the end, we came up with our own design.

brakesThe next step was to acquire two old wheelchair motors, which were 24 volt DC motors, to drive the main chassis of the MowBot.  (By the way, we named it “MowBot” in the summer, and “SnowBot” in the winter.)  We found a good pair on eBay, each with a lever attached to disengage the gears so that the MowBot could be pushed when needed.  We had to remove the brake from each motor, which wasn’t too difficult.  The lawn we are mowing is flat, so a brake is not necessary.

The frame for the chassis was constructed out of angle iron and aluminum, with 5/16″ bolts used to secure everything together.  We added a small tray on the bottom at the rear, which is used to hold the two 12 volt batteries that power the wheelchair motors.

chassisThe wheels were found at Harbor Freight, and the lug nuts for the wheels lined up nicely with the holes in the wheel hubs so no modifications were needed.  Each wheel can hold up to 300 pounds, and we had no trouble even with the heavy weight of the chassis plus the electric lawn mower.

The Neuton mower was mounted with some 1-inch aluminum bars that were hung from the chassis sides.  We bent them slightly with a vice so that we could stick the Neuton axles through holes we drilled in the aluminum.  This way we could use the original height adjustment lever on the Neuton mower to raise and lower the blade height.  mower mountWe found out later that even with everything carefully measured to be exactly the same height as the original Neuton wheels, the mower height was still too low and we had to raise it when we first tested out the MowBot.

electronics box

After the batteries were installed on the back of the chassis in two appropriately-sized battery boxes, we mounted an electronics box on top which would contain all of the electronics necessary to control the MowBot.  With the batteries mounted so low, the MowBot has the right weight distribution and center of gravity to allow for easy turning, even with the Neuton mower attached.  The electronics box was designed to be waterproof, which is good since everything seems to collect grass clippings when mowing.  Keeping these out of the electronics was a good idea.

In the electronics box, we used a Sabertooth dual 25 amp motor driver (2×25) from Dimension Engineering to connect to the receiver from the remote control, the batteries, and the motors.  The connectors you see in the middle are simply two homemade bus bars since there are numerous connections to the two 12 volt main batteries.

side switch

A PicoSwitch radio controlled relay, also from Dimension Engineering listens to an on/off command from channel 5 from the RC radio and tells an Arduino Uno to turn on the Neuton mower.  An output pin from the Arduino turns on a solid state relay, which turns on a continuous duty solenoid, thus turning on the Neuton mower’s power.  The Neuton mower’s power comes from a separate 24-volt battery that fits in the Neuton mower.  This means that we can remove the Neuton mower in the winter and attach a snow plow or electric snowblower!  To power the Arduino with ~8 volts DC, a voltage regulator from Dimension Engineering was used, which can take the 24 volts from the drive batteries and convert it into a lower voltage the Arduino can use. A switch was installed on the side to turn the power on and off.  Since the main batteries are used to engage the continuous duty solenoid, with the main power switch off, the Neuton Mower is also turned off.  Finally two voltage meters were installed with a DPST on/off switch to show the condition of the batteries.

voltage meters

Although many of the devices had heat sinks attached, we didn’t expect them to heat up much in the electronics box.  However, after our first attempt to mow the lawn with just RC control (~45 minutes long), the Sabertooth motor controller was quite hot.  The other components were simply warm to the touch but not hot.  Going forward, we will need to install a 24 volt fan to pass some air over the Sabertooth’s heat sink.

The completed mower is fully functional, but needed a better paint job.  Because we expect to use this SnowBot in the winter, the steel portion of the frame must be painted to resist rust.  Thus, after we had completed our full assembly, we disassembled everything and gave it a custom paint scheme.  After a coat of primer, some black paint, and then some green paint, it looked much better.  Gone were the rust spots from the steel angle iron.  Plus, it has much better visibility, which is important for a robot with a spinning blade which might be coming at you!unpainted mower

The next step in this project will be to add an APM 2.5 autopilot (an APM:Rover), which will contain a GPS and inertial components necessary to allow us to turn this loose in an autonomous mode.  To allow the operator to steer it correctly in RC mode, we may also add a video camera to the front of the mower.  Eventually, we may need lights installed for low light conditions, such as a snowstorm or mowing near dusk.  However, we couldn’t wait to post this summary to show off the functional MowBot/SnowBot.  It was a fun project!

Here is a video of it in action: MowBot Video



14 comments on “Autonomous (and Remote Controlled) Lawn Mower

  • Ed

    How did you put together the component for the remote control start? Can you send a parts list with what capabilities your starter has. I have briefly thought of starting a Gx 390 Honda engine from about 200 ft away.

  • Fritz

    Nicely done I am in the middle of making one with a gas mower

    I have what looks like the same wheelchair motors
    where did you find the hubs to put on them for the tires you used? The stock wheelchair tires that came with my motor do not look up to the task


    • Alligator

      The wheels I purchased were from Harbor Freight. They happened to have the bolt patterns I needed to match the wheelchair motors already, so I didn’t have to adapt anything to make them work. After using them for a season, I’m not impressed with the quality of the wheels, so next time, I would want to try something different.

  • Fritz

    Thanks for the reply

    Did you use a diode in combination with the Switch on the Sabertooth?

    I talked to the tech at Dimension engineering and she told me if you put a switch in the circuit you need a 40-60V 5 amp diode to protect the Sabertooth since it is a regenerative circuit, they even emailed me a diagram of how it would go

    Was kind of surprised by this I would have thought they would have included that in all the
    information on the PDF of the unit

    • Alligator

      No, I don’t believe I used a diode. However, if they recommend one, it sounds like a good idea.

  • Jeff

    Wow, nice work!! I’m in the build process with a gas one as well. I also had the idea of making the chassis universal – except instead of the snowblower, I have a large stereo ice chest that I want to put on RC (in the deep south, we don’t get a whole lot of snow 🙂 ). Figured the lawn mower would be great in the summer – then when football/tailgate season comes around, it’ll be handsfree as well.

    But for now, I’m having difficulty with fabricating a motor bracket to attach to the frame, as my wheelchair motors have a weird crescent/triangular pattern, and all I have is simple metal working tools (basic welder, grinder). So I figured I’d just find a used frame from the wheelchair the motors were designed for and hack it up.

    Oh and instead of buying a picoswitch, there’s an instructable for hacking a normal servo motor into a picoswitch, for when you add your lights!

    Thanks for giving me some new ideas to try – and keep us posted on the GPS/autonomous feature!

    • Alligator

      Thanks! Thanks for the information about the picoswitch.

  • Bruce Lunde

    How are you doing with the autonomous mode? Are you planning some sort of sensor fusion to get the inputs to keep the unit driving relatively straight lines? What are your thoughts on lidar use in this effort for obstacle avoidance?

    • Alligator

      Version 2.0 of the Mowbot is currently in progress. The newer version uses an all new frame, much higher computational power and a better GPS. I’ll prepare a write up shortly on this once I finish some testing. LIDAR would be an excellent choice, but the known problems are (1) cost (although this is coming down); and (2) performance in bright sunlight. I’m not currently using a LIDAR unit, but I may try one if I can get one at a reasonable cost.

  • Gilby Dettori


    Great job on the first version! really curious as to how you’re getting on with Version 2?
    Planning on building one myself with ArduRover on a Pixhawk. Also very much thought of the LIDAR option(as they’re nice and cheap now).


    • Alligator

      Yes, I really want LIDAR but I haven’t found one that works well in sunlight yet.

  • phil st. romain

    i am a novice at the mowbot world.
    i bought a used jazzy.
    i was wondering why the existing controller unit couldn’t work.

    • Alligator

      It might, give it a shot and see how it goes!

Comments are closed.