Moar planning!

Date: 2021-01-13
Author: Andrew

We got together in the evening and did some more planning on Trello. Overall we have broken version 3 into three chunks.

• Add a web interface to the Beach Bot and connect through a network (e.g. personal hotspot on a phone) to be able to see state of services, sensor readings, etc. Once up and running convert the phyiscally wired functions (shutdown, take a photo) and make them software driven.
• Then start making physical changes by powering the Pi from the RC car battery and attempting to steer/throttle via the web interface.
• Design and print new components to house the Camera, Pi (and possibly an Arduino) with a solid attachment to the RC car.

We broke the web piece down into smaller chunks as there is quite a bit there to get it working.

We also need to give the RC car a test on the beach to see if it works OK. Otherwise we might need another platform. And we'll need to add a power convertor to get the voltage from the RC card into the Pi at 5v.

I was concerned about the GPIO pins running out as an H-bridge set up can use plenty, four for the GPS, four for the m6050 (gyro/accelometer/tempreature) feels like we are running low already. Rohan suggested that we use a common ground on the RC car. Besides saving pins it means that if there is a grounding problem then nothing works as opposed to possible partial fais which will be harder to debug.

Another idea was to add an Arduino to drive the pyhsical components. This has many advantages such as a simple kill switch if the Arduino loses contact with the Pi because it's crashed making it quick and easy to stop the bot. Otherwise if the Pi service controlling the throttle hangs then there may be no way to stop it! Plus it will save more pins on the Pi. Downsides include increasing the power draw, probably more heat and more physical space but sounds like a reasonable trade off.

At some point we need to start managing the SD card build as well as we are deploying more and more stuff onto the Pi and if the card dies then rebuilding from scratch will be a slow and painful task so that's also on the to do list... at the end.

Couple of things are undecided at this point besdies the Arduino. One is the battery monitoring. Should not be too hard as Rohan has created one before but need to be aware of scope creep. The other is classifying the images when they are taken. This would be good as then it's done there and then. Plus the litter audit can be auto generated at the end of the session! It's a nice to have at the moment so we'll see how we go...

So planning nearly done. Time to start getting into build!

• Next: Website is up!
• Previous: Second fail in a row