POC robot vaccum

vaccum shitbox

I wanted to think of a hardware project that would make use of my background in sofwtware/ML, i therefore went with a robot vaccum, which in my case basically made use of neither. But in reality I just wanted a project that would expose me to some foundational concepts within hardware projects, such as:

And I feel like this project has somewhat covered those bases.

The LIDAR is pointed downwards, so it thinks something is obstructing its pathway, making it spin around trying to find a way out

I stopped the project, once I knew that the "robot" was capable of reacting properly to LIDAR input, and orient itself throughout the home. Because in reality I could not care less about a robot vaccum, I want to move onto something else. 💀

List of (main) used components:

Vaccum housing

The entire frame was 3d printed. The frame was modelled, so that you could slide in the dustbin, which had a intake hole of dimension: whereas the motor intake hole was modelled to be . The floor hole is deliberately ~2.4× smaller than the motor intake. That constriction is where the air accelerates. High velocity at the floor surface is what lifts debris off the ground.

The vaccum motor itself is far more powerful than the average robot vaccum (30 - 100 watts), and that was on purpose as i assumed my poor modelling skills would lead to some air leaks, meaning I wanted more power to compensate my skill issues 😪.

Control Logic

The center of the control logic layed within the Rpi5. This then controlled the power of the vaccum motor, and the speed/rotations of the driver motors (wheels). The LIDAR relays the information over to the Rpi5 which then tells the wheels to stop, turn, go forward.

An issue is that the 12V that would be sent from the battery would be too strong for the Rpi5 to accept. Meaning we need a buck converter to reduce the voltage before hitting the Rpi5, which should sit at around 5-5.1 V, as you can see below.

Adjusting step down

Lidar

The lidar I purchased: https://www.amazon.de/dp/B07VLFGT27?ref=ppx_yo2ov_dt_b_fed_asin_title Didn't come with the advertised cables, and only jst head to female dupont ends. So I tried just using the gpio directly with the dupont ends, for a solid day, to no avail. Ordered a dupont to usb-a adapter, and slid that into the Rpi5, and it worked immediately, so keep an eye out on that, if you plan on connecting the Lidar to gpio directly...

Putting it together

As mentioned, a robot vaccum is not something I want, or interests me as a build, just used it as an exposure project. I therefore didn't want to solder my cables shorter so it would fit neatly into my frame. Which then led to a monstrosity of a build, where long cables were overlaid ontop of each other, oh well.

But basically it was just a larger frame, with enough floor space for the wheels to be screwed onto the frame, with 2× Pololu 3/8" plastic ball casters attached to the underside for stability. And a cylinder entry for the vaccum motor, with an entry under the vaccm motor where the dustbin could be slid into.