My other group members have taken the actual robot, so I don't get the pleasure of smashing it, hence I'm venting here.

Ok, a bit of background:
One of the assessable parts of a mechanical unit I'm doing this semester (TRC2100) is to design and build a submission to the Warman competition this year, run annually by University of New South Wales/Australian Defence Force Academy. Unis around Australia participate by getting teams of students to build a robot that performs a set task in accordance to a list of rules. This year's challenge in short is to create a robot that:

• climbs a 6.5cm perpendicular wall
• deposits two hollow softball-sized balls into the target pockets
  
• picks up same-sized balls at designated positions sitting on tees at random heights (up to six can be placed)
• climbs another 6.5cm perpendicular wall
• stops in the recovery area

There's a raft of rules, including:

• a score penalty if you source electrical power from anything other than up to 4 AA batteries
• vehicle must fit within a 40cm cube
• must be autonomous (cannot be controlled in any way once started)

The most important rule however, was - the vehicle must not at any point be in contact with any point outside the track, otherwise a score of 0 will be awarded.


As with all eng projects, we planned grand at the beginning - tank-like vehicle, extendable arms, LEDs, fibreglassed exterior, victory song to be played at the end - but procrastination took hold and it ended up being a fairly simple vehicle that we assembled mostly this week. It was made out of meccano and bits of wood, powered by 4xAA batteries connected to a motor and gearbox, driving tank tracks on both sides. The tank tracks helped get it over the wall. The balls were deposited as they rolled out of holders angled towards the pockets while the vehicle was descending the first wall, and the vehicle was stopped through a simple uncoiling piece of string that triggered the switch when it ended. Time constraints meant we didn't attempt to collect any of the balls.

We had to pay for everything out of our own pockets which sucked, because none of my group members had any electronic equipment (e.g. Lego Mindstorms) and sort of gave an unfair advantage to those who did or had the funds to buy it, but I digress...
 

So after working on it for days, including a near all-nighter the night before presentation (there was an annoying track alignment issue causing it to veer left), it was time to demo the newly christened "King of the Castle" (Borat reference).

I wished I filmed it, but I forgot so you'll have to deal with my description:

• we did a quick pre-flight check (there was a time-limit)
• flicked the on-switch
• it rolled along, relatively straight
• climbed over the first wall fine
• balls rolled out as it descended the first wall, one missing completely, the other rolled in after a while
• climbed over the 2nd wall fine, now with a slight angle to the left, but a still a fair way away from the left edge
• entered the recovery area
• reached the end step
• rolls over the end step, falls outside the track, before stopping on top of the end step, the front of the car touching the floor outside!

Arrgh! We were so hopeful after it climbed the last wall because we had accumulated a decent amount of points (many cars didn't even make it over the first wall) and the stopping mechanism was well tested before. So from a relatively high score, we fell all the way back to zero (see rule above). There goes most of my 20% this project was worth of my final unit mark. All the time, money and sweat wasted.

The second run stuffed up due to an unforeseen technical mishap (the tank tracks snapped at a point of connection) while trying to get over the second wall, so we ended up with that score but with no balls going into the pockets this time, it was a relatively low score.

So much for being "King of the Castle". What probably happened was the coil unwound just that tiny bit while we were waiting to demo, hence giving more distance for the string to run before triggering the off switch. Ok, so maybe we could've avoided it if we had checked the string during the pre-flight check, but hey, it's a lot easier blaming a vehicle than ourselves. We weren't the only ones with bad luck either - there was an awesome vehicle with electronically-controlled arms that folded out to deposit the balls at the precise location, then folded in again, and a 'mouth' at the front that collected balls in front of it, but it failed to get over the first wall twice. Another electronically-controlled one malfunctioned as well even though it was working perfectly during test runs the night before. Other awesome ideas included machined notched wheels that straddled the walls completely, crossing the wall like it wasn't there. There were some very successful ones though, some entirely mechanical, but it was all luck basically because I don't think any could replicate its success consistently over the two runs.

And that's why robots are evil. Remind me why I'm studying mechatronics again?