Back to KEITH Evolution Blog Archive

Reflections on the Development of a Robot

In our first blog post back in September we outlined out thoughts on how Keith Evo might work out. In general we have stuck to our principles and the summer research paid off. The major change to what we thought we would do is a lack of attachments. We have just one this year which is to attach the wire for Pi Noon. Everything else is built in to the chassis. This came about because while having attachments last year was fun, one of the problems we suffered was lack of time to swap things over and difficulties with plugging things in and then remaining reliable.The timetable this year looks to have sorted the lack of time out, but we erred on the side of safety and KEITH Evo will hopefully be reliable.

The one prediction that was totally accurate was that KEITH Evo would involve servos, bent wire and rubber bands. We’ve added copper pipe, steel pipe and structural soldering to that this year!

Once again we have been delighted with the support from various quarters. Leeds Hackspace allowed us to purchase components at silly prices, laser cut one of the chassis prototypes for us and various members spoke to Harry unintelligibly (to me!) about coding and good practices. He seems to have taken that to heart and has a file some 500 lines long for the challenges for PiWars which even I can follow… The DT team at Harry’s school have provided laser cutting for the original prototype and the final article at no cost. Harry will be taking KEITH Evo into school after PiWars for ‘show-and-tell’ as payment!

Better motors have made a huge difference. KEITH Evo is better balanced as a result and has power a plenty. He could have been faster, but experimenting would have required a bigger financial outlay, so we have stuck with our first choice. As with all of these things, there is a balance to be struck and we are happy with what we have. We eventually picked the Vex Robotics track kit for KEITH Evo. I thought (and still think) that the Vex stuff is ridiculously expensive for what it is, but the tracks and wheels are good quality and should last. Hacking (in every sense of the word) the wheels to mount to our motors took a deal of courage, but it did work. The tracks would have been perfect on short carpet, but they slide too much on a smooth floor. The final resolution, having tried lots of ways to improve grip, was to make good use of a collection of loombands. Possibly a world first!

Loom bands!

Home made circuit boards… Last year Keith had a Pirocon motor controller board at his heart, mounted directly onto the Pi. That board did lots besides motor control so much of this year’s work has been working out what we needed to do instead. The only major issue we had was the release of magic smoke from the first motor control IC. A realisation of the differences between stall currents and the current when running free directed Harry to redesign the board using new ICs. A small voltage regulator and a variety of resistors allowed us to provide 5v feed to components with 3.3v control signals to/from the Pi. We’ve added transistors to control the switching of LEDs as well.

So what have been the biggest challenges? It’s difficult to say, but in terms of software the three point turn is the hardest. We thought we had it sorted last year only to have things go wrong on the day when Keith did not behave as expected. Salt was rubbed into that wound when Keith performed the three point turn faultlessly (see YouTube!) after we had used our available rescues. I’m not going to tempt fate, but KEITH Evo is now repeating the test consistently after a number of hours of refinement. In terms of hardware, the biggest challenge has been incorporating the skittle challenge cannon into the chassis with enough punch to be effective without damaging the chassis when it goes off! Reloading is not perfect, but it works… We have had two balls to practice with – one too light and one too heavy. KEITH Evo manages well with the heavy one, so it should work with care!

Skills in fault finding have been developed. A home-made circuit board that has worked fine but then does not after a new component or wires have been added needs the magnifying glass to find that one wire that has lost its solder (over enthusiastic soldering on my part!) and has produced a dodgy connection as a result. Very frustrating.

I have really enjoyed the exchanges between some teams on Twitter. Encouragement, goading, mild abuse. It has all been very good fun during the build and I am really looking forward to meeting them in person and seeing what they have produced. There is no doubt that once again PiWars has entrants with a huge range of abilities, skills and knowledge. All are enthusiastic and whatever the outcome of the competition we will all have a great day, meet new friends and remember it for a long time to come. Thanks to Michael and Tim for all of their hard work in organising everything!

Writing as the dad of a geeky teenager, PiWars has provided a fantastic focus for his computing/coding skills and aspirations. He may still have spent hours on the computer researching code, components, circuit designs and whatever else, but the final result is real and meets a clear design brief. For my part, it has been a great excuse to get really involved in what makes him tick. I have no knowledge of coding (except a bit of Basic from the days of the ZX Spectrum… ) and little or no knowledge of electronics. I can use AutoCad (normally for mapping and civil engineering drawings) and I have a love of model-making and can wield a soldering iron(sort of). The combination of the two sets of talent has produced two working robots, so far. We’ve had our fall-outs and frustrations, but I am very proud of Harry and the way in which he has worked through the problems to produce KEITH Evo.

At this point last week, I had decided that if the raft of problems we were experiencing had not been resolved by today then there would be little point in making the 320mile round journey from Leeds to Cambridge. Some hard work this week (wrapped around the day job and homework) meant that we finished at teatime today happy that we have a working robot and that we have a few evenings this week to practice and refine a few bits. Whatever happens on Saturday, I will be happy that we gave it our best shot.

We will update this blog after the competition and then Harry can concentrate on his A levels, EPQ and whatever else he wants to do. I quite fancy building a drone…

~Andy


If you haven't already, don't forget to follow us on Twitter @KEITHtheRobot for updates on when new blog posts are live!


Back to KEITH Evolution Blog Archive