Progress is somewhat slow. It's hard to see how much research need to be done, as each piece of research leads to two more pieces being required. Hopefully, I can reach the edges of this knownledge "tree" in the near future.
I'm pleased with how the navigation softare worked out, and had the opportunity to test it on the HRSFC robotics team's (Systemetric) robot. It worked very well, and hopefully will work on the much smaller robocup robot. I didn't get a chance to write the control loop code, as Systemetric's robot did not have a compass.
It's been very hard to find appropriate wheels, as the type I want don't seem to exist on the internet, and the ones that are similar don't seem compatible with any suitable motors. I would quite like to have a go at custom manufacturing them.
I do feel that I may be falling behind a little. It's frustrating not having any parts to experiment with.
I feel progress went really during well these months. Going to the robocup UK finals provided plenty of information about the design of other team's robots.
I fear the PCB route may be an expensive and time consuming dead end. The online PCB services make it nightmarish to get quotes, the pricing is unclear, and above all, I don't really know what I'm doing. On balance, I think I'll switch over to using stripboard, since it's cheaper, modifyable, and faster.
The wheel solution I came up with looks like it's going to work. After a suggestion from my teammate to use split-rings from keychains as the axles, the degin became obvious. Additionally, it looks affordable. I feel the design process moved along fairly quickly, which I'm pleased with. Hopefully, it can be ordered fairly painlessly, and all the parts will just work together.
Getting the shopping list mostly complete is a big plus. Most of the parts have been ordered, and will hopefully arrive soon. When that happens, the design process can stop being theoretical.
This is where everything started coming together. The parts arrived fairly rapidly after ordering, allowing us to start building the robots
It was nice to finally have the FEZ Panda II at hand, as it meant I could learn how to program them the easy way - by doing it and seeing what goes wrong. The tinyCLR forum was very useful, and greatly improved my grasp of the library for the controller. The main problem I had was that the official documentation for the .NET Micro Framework is quite light on information, and fairly vague
The over sensitivity of the IR sensors is very worrying. In order to be useful, they need to have a very directional signal. Without shielding, the sensors pick up the same signal wherever the source is, provided the distance is constant. Hopefully, I can come up with a solution soon. This really is critical to the final project
To finish off the month, the custom-machined wheel parts arrived. These fitted the split rings and washers perfectly, which was very pleasing. They were unfortunately nightmarish to assemble, but I am very pleased with the end products.
This was a busy month. Pretty much all off the building and programming happened in this month, along with small design tweaks. Only half way through the month could all the remaining parts be ordered. I do feel it was cut a bit fine.
The IR sensor shielding was finally solved adequately by painting the sensors in matte black paint. It was a real relief to have this problem out of the way. Additionally, the sensor aggregation worked perfectly.
I think the most successful progress this month was with the native RLP code. It cut the execution time down from around 100ms to only 10ms. That's a substantial speed boost. The accuracy of the ultrasonic readings was also improved
Making the stripboards was a new experience for me. I'd never worked with the stuff before, nor had I done that much soldering in one sitting before! It was really frustrating not to have wires to experiment with sooner, but the shipping just took longer than expected.
The "broken" motor drivers were a worrying setback. I checked and double checked the circuit diagram, and could see no differences between the broken and working boards. The problem was eventually found to be short-circuiting bits of drilled copper, though exhaustive connectivity testing.
It was good to see the robot finally driving around, after 5 months of theory. It was about as fast as expected, and the software worked after a few bug fixes.
The first real problem came right at the end of the month when one of the motors stopped working. This was very bad for two reasons. The first is that there was very little time left till the competition. The second is that we had no idea what caused the breakage. The software was changed to prevent rapid changes of motor direction, but it is unclear whether this will prevent further breakages