Pi-Wars Weekend is Imminent!

Only got about a week to go before Pi Wars 2024!  The organisers are sending piles of organisational stuff, including timetables and logistics.  We are in a frenzy working on last minute stuff.  One does wonder how the organisers are managing to get the mass of stuff they need to do done!

Anyway, thought we would provide a 'mopping up' blog post, summarising our general readiness...

Lava Palava

This was supposed to be an easy one.  We developed some simple line following code early on, that would use signals from 2 'line follower' sensors to follow the white line in the middle of the Lava Palava course.  However, the 'hump' in the course has thrown a potential spanner in the works.  We have lengthened the wheelbase to improve stability and guard against Dan-ED 'toppling over'.  But there is an obvious danger that the line sensors might get 'knocked off' by the hump.  We may need to add in signals from the ultrasonic distance sensors as a backup...

You can see Dan-ED following the white line using its line follower sensors.  It generally works well for shallow curves and bends up to 135 degrees but not 90 degrees!

Escape Route (Blind Maze)

This one is pretty solid (see our 'How to Navigate a Maze!' post).  We will be running this in purely 'Autonomous mode'.  We did originally incorporate some machine vision code detecting the colour of the walls to guide Dan-ED but, in the end, we were able to implement much simpler code just using feedback from the 3 ultrasonic distance sensors and some clever decision trees to guide Dan-ED. It works pretty well.

Here's a vid of Danny driving Dan-ED through the maze...

In this clip Dan-ED is navigating the maze 'autonomously'.  Clearly the machine can do better!  It takes about 25 secs to complete the course.

Minesweeper

We are almost there, and should be able to do this 'autonomously'.  Our machine vision code can detect the lit squares (see 'Blob Tracking' post).  The tracking code can then steer Dan-ED to the coloured square.  It works quite well.  We've managed to implement some additional code for Dan-ED to 'stop for 1 second' when it reaches the square before going on to hunt for the next lit square.

The Zombie Apocalypse

We've been working on this for a long while.  The CO2 powered nerf shooter concept was developed for the original Eco Disaster challenge back in 2019.  It still works.  We have improved the pan and tilt mechanism to steer the nerf shooter more finely.  This allows for better aiming accuracy.

Being made of lego, it was relatively simple to do.  Ben built the original pan and tilt.  He is at Uni now, but was back over the Easter to 'modify' it for us.  He has basically added additional gearing stages to allow the pan and tilt mechanism to steer more smoothly.

I'll let Ben talk more about the nerf shooter and what he did to improve it.

Here's the whole shooter assembly mounted on Dan-ED.

Jerry and Danny setting up and putting the shooter into 'action'.  It's pretty accurate and should be able to knock down the zombies wherever they might be hiding!

Eco Disaster

We are still debating whether to go 'Autonomous' or drive Dan-ED 'manually' using the game pad controller.  The machine vision code has been progressing well and I think we could get it to work.  

Here's the rub... if we do it manually, we know it can be done reliably and quickly, but 'autonomously' it will definitely be a lot slower, purely because we have to use a smaller pickup that can only pick up and move one barrel at a time.

Autonomous also brings a high risk of something going wrong.  The machine vision code guiding Dan-ED is highly tuned to lighting conditions.  We have added a 'selfie' light to try and mitigate against varying lighting but it is not perfect.

You can see Ben driving Dan-ED to 'manually' to pick up the barrels.  He can easily do all 6 barrels in about 35 secs.

Notice 'cloud city' Boba Fett is also in the driving seat to help steer Dan-ED!

There is a different 'wider' pick up assembly for 'manual' mode.  It can pick up two barrels at a time which speeds up the challenge.

I think we will have to decide on the day whether to go 'manual' or 'autonomous'.  From purely a points scoring perspective, there isn't a huge advantage going autonomous.  We need to weigh up scoring 100% doing it 'manually' against maybe 50% in 'autonomous' mode .  Even with the higher 'per barrel' points available for 'autonomous' we may score more by achieving 100% in 'manual' mode...