In Programming the ultrasonics to work with mecanum wheels for the autonomous ‘The Canyons of Mars’ challenge (a maze), we overthought the process. The mecanum wheels – by allowing the robot to strafe – simplified the code significantly as we no longer needed to calculate turn angles at the corners. Thus, we invested more time into making the logic behind the Ultrasonic sensors as fast and efficient as possible. It was a mistake.

What we had forgotten to take into account was that, even though the mecanum wheels increased our manoeuvrability, they would still be limited by the inaccuracies and slowness of ultrasonic sensors. This meant our early experiments resulted in the robot strafing too early due to a (annoyingly rather common) anomalous spike in the readings from time to time – the Ultrasonic script was too efficient for its own good, being far too sensitive and resulting in minor blips in the ultrasonic readings leading to the robot quite literally going in circles.

Thankfully, after using print statements everywhere to debug our script step by step, we eventually managed to calibrate the sensitivity in such a way that it only took a turn when the values had stabilized – i.e. stopping at each corner to give the sensors time to catch up etc.

Next year we will definitely be investing in infrared proximity sensors (far more accurate) to allow fast and smooth movements around autonomous challenges.

Frostmite is a complicated robot with many cables. However, unfortunately when we were building frostmite the instructions were wrong which caused us to incorrectly wire the thunderborg and the raspberry pi to the power supply.

This resulted in us breaking the raspberry pi and both motor controllers. After we got another pi and motor controllers, we started to put frostmite back together piece by piece. After we had connected a piece of the robot, we triple checked it to make sure that it was right and that nothing would break before we moved onto the next piece.

Once the robot was assembled with all the new components and wires, we checked all the wires again to make sure that it was all correct and then we turned it one. Luckily, when we turned it on it all worked and nothing broke.

Next year, we will consider labelling our cables to make sure that they are connected to the right component.

Despite our overall success last year, especially in the obstacle course and autonomous maze challenge, we were less successful in PiNoon (aim to pop the opponent’s balloons first) and lost our balloons rather too quickly, thanks to the rival team’s small, agile robot.  To humiliate us further, the enemy was built by a team of primary school children so the defeat was mildly embarrassing, to say the least.

Determined to avoid a repeat of last year, we’ve invested in a set of brand new Mecanum wheels which, when used together, can move the robot from side to side as well as forwards and backwards, giving us the flexibility we lacked. The wheels consists of a set of 12 rollers at the circumference of the wheel, positioned at 45° to the centre, as pictured below.

Image result for mecanum wheel

In order to integrate these Mecanum wheels with Frostmite we had to purchase an additional Thunderborg which we used in conjunction with our existing Thunderborg and Ultraborg via a daisy chain arrangement. This didn’t go amazingly at first but once we got to grips with the cabling we were able to start writing the motor code, which is of course vital as all the challenges require the robot to move.

View our PiBorg forum post here:

Results from PiWars 2018: