I took a regular old Roomba, squeezed
some high performance motors, electronics and RC gear inside, and then hit the road
to see if I could set a new record for the world's fastest Roomba. This isn't my first Roomba rodeo,
and I know a bit about electric RC. But just how fast can a Roomba go? There are a handful of videos
claiming to be the fastest Roomba. But I soon found that this isn't a Roomba. And this is definitely not a Roomba. And then I came across this one, and this is the real deal. I also found a story that covers
what's inside. It's got bigger motors
paired with planetary gearboxes. I estimate that it's good
for about 13 kilometers an hour. I’m considering this the benchmark to beat. And I have a concept in mind to achieve it. And to get there,
I'm following these four design rules. It has to look like a Roomba. Use two drive wheels. With one motor per wheel. And it has to suck. It will be challenging to fit everything
inside the cramped Roomba chassis. And there form isn't optimal
for aerodynamics or handling. We need to find a Roomba that's up to the task. I started
by removing all the parts from the Roomba. This doesn't take too
long as they have a very modular design. I was left with a pile of bits
that, as you would expect,
was pretty filthy from vacuum duties. So I cleaned them up before moving on I needed to make some serious space
in the chassis to fit the new drive
train and electronics. This was a lengthy and smelly process
and I was careful to leave enough strength in the chassis. I went with brushed 775 motors. That's spin up to about 20,000 rpm. These will transfer power
to a super strong modular planetary gearbox. To start with
though I’ll just be using a 3 to 1 gear ratio, which is the longest available. There's quite a size difference
between the original and new arrangements. And of course there's one for each wheel in order to fit everything in the right
spot. A 3D printed a dummy wheel, which is quite a bit larger than the stock
one to get my fitment right. The wheel design has a positive offset
in order to fit over the gearbox and stay within the bounds of the chassis. I printed some plates in PETG that will attach the gearbox
to the chassis and checked
clearance before installing everything with everything fitting nicely. I printed a pair of new wheels in PETG. I used TPU filament to print some tires and fitted them with epoxy. I'm using one motor controller per side and they can handle about 60
amps of power each. I picked up most of these parts at Andy
Mark, who were very helpful. Anticipating possible
cooling problems down the track, I made up some DIY adhesive heating
compound and attached heatsinks to the motors. I soldered it all together
and decided to ditch the heavy nickel metal hydride battery
in favor of two three cell LiPos. I was pretty restricted with what I could actually fit in there. I could now power it up to program
the motor controllers to suit the motors
and then give it a test on the bench. I printed a few little accent pieces and
then put the whole thing back together. I didn't know what to expect,
so I gave it a little run on the workshop floor to start with. I'm glad I didn't have any electronics mounted in that dust collector. I added some wheelie protection and a Bluetooth GPS
so that we can measure its speed. The tires provided next to no grip and that caught surface was very grippy. So I found some urethane rubber to cast
some tires using a new wheel design
and a two part mold. I added a few drops of dye and then poured the urethane. I left them to sit overnight
and to my surprise, they came out of the molds easily. Not bad for my first go, but there was still some room for improvement. Next up, I needed to get those wheelies under control and to improve the handling. I added some different
exponential settings and you may have noticed
that the Roomba doesn't suck yet. So using the existing power rails,
I wired up the rear sucker and front
sweeper to a switch on the receiver Seed studios, kindly sent me
one of their Xiao ESP32s so I wired it up to some LEDs
for some night effects. Someone must have known I was coming. One broken gearbox mount and two broken wheels. The urethane tires are extremely strong though. Here you can see the impact
that broke the gearbox mount and a wheel, and then the lateral force
that probably broke the other wheel. Time for some new tires and stronger
wheels. I didn't get the molding
quite right, again. And four cell LiPos that will deliver more power To make room for them, I quickly realized
this was a motors out job. I still wasn't happy
with the wheelie control, so I made a little closed mold
for the wheel, which worked wheely well. And I pilfered some bearings from an old drone motor
to reduce friction. And then I had some electronic problems with some intermittent pulsing
that would kill the drive to the wheels. I suspected that I was running
too many amps for the motor controllers, but looking at the software,
there was nothing unusual going on. I eventually attributed the problem to the onboard five volt
supply from the motor controllers. I think. So I decided to power the receiver with a step down voltage converter
instead, which I tucked away in the back. And now there is literally
no more available space on the chassis. The extra power had again worsened
the handling of the Roomba. It was a real handful to control, but it was faster
after only the first run. The second run was faster again, but that was the fastest
I could manage before being forced to retire again. We've got damage. We've got pretty similar damage to last time
despite the stronger wheels. I think this could be a good candidate for some machined
or 3D printed aluminum parts. Or maybe I'll detune it a little
for some track time. That could be fun. I’m really pleased
with how this project turned out. But it may not be the end just yet. Let me know in the comments
if you have any ideas for me.
