#mattblog

Finally, the time has come! The epic conclusion to our micro mouse tale! Greganog Hall! I can’t believe it’s all over, after a whole year of working on Clyde, the assessment is done and we can put down our tools, log off our computers and admire our handy work. The trip to Greganog was a blast, with plenty of ups and downs. The bus being two hours late to get us there was not a good start, especially as I was eager to get there and get to work putting the final touches on Clyde. But once the bus did turn up it got us to our destination safe and sound. After checking into our rooms and dumping our bags, we were treated to the first of many very tasty meals, and then it was down to work! Boy did I not know what was coming to me. We had a fairly easy beginning, until I realised the laptop that I had bought with me had no code editing software of any kind, which meant I would have to edit all code on notepad without the benefit of debugging, which was somewhat unsettling but I came to terms with it and we continued to move on at a swift pace. We had gone to Greganog with set goals. Ben and Meg would work on presentation and documentation, while me, Soren and Tom worked on any tweaking and testing that would be required. I also had a set of coding goals in my mind. First of all was to finish combat, and then move onto tweaking, especially with the line following improvement code, which had been plaguing me for weeks. So within an hour or so I had the combat code adjusted, and some tests with the team had confirmed that it was ready to go! Or so we thought….. more on that later. So, on I went to the line following code. As mentioned in previous blogs I had been trying to smooth out the line following code for weeks on end, with no success. However, I felt I was getting closer and closer to the solution so continued to work on it. This had me tearing my hair out for many hours and took up most of the afternoon. When it got into the evening the rest of my team had gone off elsewhere to complete other tasks, and I decided to test Clyde’s new code by myself. I armed myself with pen and paper and took Clyde to the *RAISED THREE FEET OFF THE FLOOR* test track, and as he was running around the circle my new code as it had so many times before failed. However, unlike the many times before I did not have a team of catchers around the table, and with my hands full and on the other side of the table I was not able to make it in time. Clyde crashed to the floor, and all the supports holding his modules together snapped. To say my heart dropped was an understatement. I picked up his battered and broken body, ran it back to our work station and tried the power switch. No life. Nothing. At all. I went and retrieved Soren to tell him what had happened and to help me try and save our project. Luckily after a few tense minutes, some replaced supports and switches returned to the on positions Clyde’s eyes shone bright again and he was back to perfect working order. This did not deter me from my testing however I did assemble a team of catchers from then on every time Clyde went anywhere near a table. Testing continued until it came to about 1:15 – 1:30, and the team assembled saying they wanted to go to bed. I too was getting tired so I set out a final set of tests that we could run just in case we had missed something to make sure we were all ready for the final challenges the next day. We were to test all the modes functions correctly and were ready to go! Sounds simple enough. The first test was obstacle avoidance. This test ran smoothly enough, the IR alignment was checked and everything was fine and we were happy with it. The next test was line following. Sadly, even after many many hours of work on this mode, the second iteration was just not up to standard. No matter what crazy tricks I tried to pull it was just not possible, with my skills anyway, to make the mode any better with two sensors. So the group voted to return to the original code. I begrudgingly accepted this and just came to terms with the fact that despite all my work the line following would not be the best (or would it? Again more on that later). And onto the final test, combat. Again we expected this test to run smoothly as this mode was completed to our knowledge. However, for this test we decided we would battle another couple of mice to test in the real situation that our code would hold up. Safe to say it didn’t, in all the practice fights, Clyde would be pushed into a corner and when his special edge mode activated he would then drive straight out of the arena! We had thought this bug unlikely but in practice it was a very frequent occurrence. So, it is now 1:45am or so, and we have a broken combat mode. The team then voted reluctantly that we would have to fix this problem. So I rushed back to the laptop and started bashing out code. Several iterations later it came to 3:00am, and Clyde would no longer fight. He would just sit in the ring and do nothing! We all decided bed was needed, and I said I would be up early in the morning to fix it. And early I was. After getting into bed at 3:30, 4 hours 15 minutes sleep later I was out of bed and dressed ready for breakfast. As soon as I was finished eating I was back into the main hall and coding. After coding so tired the night before I decided what I had written was probably useless, so I scrapped it and started with the previous buggy version and worked from there. Having had some sleep, I made much better progress and by the time the rest of the team had joined me in the hall I had a new working version with the bug fixed and some other safety measures in place just in case. So with Clyde ready to go we charged the batteries and waited for the final assessments. After lots of watching and waiting for what felt like forever, and some tea and cakes, our first test was upon us. Obstacle Avoidance. This was a nice first test I thought being one of our better modes. The test went very smoothly with Clyde avoiding nearly everything with only IR and using his bumpers only on occasion. The judges seemed impressed and were also interested by all the 3D printer parts of our mouse. The second test which came in rapid succession was combat seeding. Some of Clyde’s issues like his love for corners came up in this test, but his good features also came through meaning we were ended up being seeded 3rd. The third test was our technical inspection which I would say our team knocked out the park! Everyone chipped in with good explanations of Clyde’s various components. And we had an ample number of special features to tell the judges about. We also managed to answer all of those questions which was very reassuring. The comment was also made that we were the tightest nit team that had come in to be judged, referring to everything as ‘we did’ this not I did and things like that. Finally came line following. This test I was not sure about seeing as were using what I had counted as non satisfactory code. The test did go well however and the team explained in depth about our method and calibration, and about the mode that I had worked on and its improvements but why it was not able to be show cased in them. The super reliable mode which very very rarely failed of course managed to fail once in this test which we all thought was a shame but it was unlucky and there was nothing we could do about it, so overall I was very happy with the outcome of this test and I think the team were too. Our final test for the day was the presentation and again we SMASHED it. So much credit has to be given to Ben on this for making such an amazing presentation, it was so good! But also credit to my team as they all did an amazing job with the presentation. In between coding the night before we had fit in two or so hours of presentation and practice, and had done some practice while waiting for the tests to start so we all knew what we had to do. We carried it out to near perfection, one of the only stumbling blocks being me and someone else being un nerved by the judge’s reaction to the ethics part of the presentation, with heads being shaken etc. We did however find out later on that this was due to the hard hitting nature of our ethical issues and nothing to do with the quality of our presentation which is a relief. We were also able to answer all the questions apart from the final one about cost very well. We also answered the questions well as a team as much as possible. We got several very good peer reviews on our presentation and good feedback from the judges, so again hats off to Ben and Meg for putting the work in to make that happen! And finally the day is done and we are on to combat! The matches themselves were a blur and it would take forever to go into too much detail, however we left the group staged having lost only 1 round in 24 to a judge’s decision, and we had the best record of either league so we were feeling good, but very nervous also. When the finals came around we were buzzing, however I was very very nervous. Not sure why but it was really getting to me! We had a convincing with in the semi-finals against delta putting us through to the final 2. The final was a close fought match, but we took the game straight 3-0, meaning we only lost 1 round in the whole tournament. We were so happy that we had done so well in fighting after putting so much time into the code the night and morning before and it was great to see Clyde pull through for the win! So the award ceremonies begun and Team Echo were awarded the main trophy and our own trophies each to take home. I was also asked to give a speech, which was terrible but I was a little overwhelmed and didn’t really know what to say, which I imagine you can tell from this blog doesn’t happen very often. We had only just sat down when we were called up to accept the award for the best…. Line following? None of the team were expecting this after we had resorted to the original code, and I was somewhat amazed at how much time I had wasted trying to improve it. Of course there were no complaints and we were all very happy have won another one of the categories. And after both these wins I was very happy to see the code I had put so much time into had pulled through again to make Clyde the best of the bunch in two categories. So after a night of celebrations a tired day of guest lecturers and returning home, it was all done and Micro mouse was over. I have to say at the start of this project I did not think I would enjoy it as much as I have, and it became more of a labour of love working on Clyde, rather than painful assignment work. And I was genuinely sad to see him going in his box for the final time and thinking we would not be working on him anymore. However, it is nice seeing him displayed in labs as a reminder of our triumphs and what our work is capable of. Doing this assignment has also given me a passion for coding and micro controllers, especially after seeing my code do so well in this assignment. So thanks you to the rest of Team Echo for being such a great team to manage, and I think we are all much better friends now. Thanks to my team for all their great working, great laughs, and the support when coding was being maddening. And thanks to Dr Davies, Mr Moody, Mr Gareth, Augustine and everyone who helped us to make Clyde as great as he is. These people spent a lot of time, pulled personal favours for us and all sorts to help us out and I thank them for it. So that’s it, Micro mouse 2016 done. It’s been a blast (who thought that would ever come out of my mouth about uni work), now onto exams and far less fun things. Team Echo Matt, out!

April started early for team echo as we decided we would come back a week before everyone else to get lots of extra work in, we were in 10-5 onwards every day to get plenty of extra hours adding feature and ironing out issues, this meant we were getting a weeks’ worth of hours extra a day so we were able to get a lot done, I think we were all glad we came back early. One issue that make us glad of the extra time was the failure of our microprocessor. And the end of term before Easter week in the last few hours we started having some weird issues with code that seemed to have been working fine before. We just put this down to weird electrical weirdness and thought it would be okay when we got back after Easter, but when we return it was not the case. As we tested things seemed to get worse and worse, which was stressful for everyone. With the advice of the demonstrators and lab staff we began removing modules one by one from our micro controller. This confirmed that none of our modules seemed at fault, and sure enough when we reconnected all of our devices to another micro controller and things worked perfectly. So after two days now that we had confirmed the problem was the board we were supplied with a new chip, however this did not fix the problem. So the whole board was exchanged for a new one and the code again worked great! We were glad to have the issue fixed but it had taken up two whole days of our five days extra. The next step we took when things were up and running was the module power, after taking all the modules off we realised we had a lot of parts to power and a lot of cables to run throughout the mouse. We decided we would tackle this problem with a transistor switch board. This would allow us to add a common earth and 5V to reduce cable clutter to just the lower level, and the transistor aspect would allow us to software control the power supplied to various modules and save energy. We got this working in the end and did some cable management on Clyde. We had to add some extra common earths for logistics reasons but it worked out well in the end. Re-wiring itself caused some issues as we had some problems with things such as tachometers being plugged in the wrong way round and causing issues. The final success of the weeks was getting some of the line following done. We wrote up the previous calibration code as mentioned in the last vlog before Easter to have coloured calibration modes to improve user experience, and we also got the first tests of white line following which went reasonably well with a few bugs and issues but it was a start. So overall a very successful week and we were glad of the time! After that week it was back to normal for us however we did start adding extra sessions each week to get more work done, to the point that in this last week we have been in the labs with Clyde every day almost. This does mean a lot has been done however. We have gone back to obstacle avoidance to fix some issues. I have changes the size of the turn when avoiding as the previous turn was not enough and seemed very prone to getting stuck. I have also added a function so if the mouse appears to be stuck in an avoidance loop that Clyde will un stick himself and move on. There has been a lot of work on line following code which has been very frustrating and is still on going. I got it working to a point but it was a little bit jagged, however this mode is very reliable. I have since moved on to trying to smooth out the line following, which has made the mode much less reliable but smooth, and no matter what I try I don’t seem to be able to make the new mode 100% reliable, but I shall persevere! I have also got near final completion on the combat mode. There were some initial issues with the sensors not want to play ball sensing white instead of black but this turned out to be a coding issues and once the line follower mapping function had been re written it worked like a charm! So I got a lot of the combat tactics coded in so that Clyde chases other mice down and pushed them out the ring. We did have some problems with Clyde being in a loop and reversing out the ring and I have fixed that for most cases as far as I am aware but need to make sure that the loop holes are closed for all cases. Our shell has also been printed finally!! We had some issues with material compatibility and the printer playing up but with the help of Mr Moody we were able to get a job print done and we finally had a shell! With a catch. The shell has been scaled to 98% if its original size which doesn’t sound like a lot but it was enough to cause a problem. This means that I had to borrow a dremel and make the rings that make up Clydes chassis smaller to make the shell fit while Tom took off some plastic on the inside of the shell, and at the end it is a very snug fit and looks great! I also added to flat sides to the IR board and some indents to allow for cable management We are now on the final stages of just going through each mode and seeing how it works and getting the groups opinions on the mode and making improvements, and we shall continue to go through this cycle until the tests to ensure that each mode is as best as it can be! So that has been April, its been a busy one! But we have come a long way and I am happy with how we have done over the course of thing project, the team has been great and we have done really well! We have some time on Monday for some last minute tweaks and then the assessments. See you on the other side, probably very hungover!

This month we started another piece of hardware which is our LED special feature. We have made a demo board that we will use to test the feature and allow us to being programming it in without having the final board that will be able to fit in the shell. I have also done a lot of coding work this month, including the code for the LED feature. I have also got the code working for the bumpers so that it is how we like it and works very nicely. The IR code is also working well and we have declared the obstacle avoidance section of the code to be completed for now. This means that I moved on to the line following mode next. So far I have coded the majority of the line following codes calibration, as we decided as a group we would like the thresh holds for the line following to be set by code each time it is used to make sure it is as accurate as possible. The calibration may have to be modified and added to depending on our needs later on but for now it is working well. The final piece of coding is switches. We decided as we are now switching modes regularly for testing reasons and that both modes are in the same code we would like to be able to switch modes without the need for re programming. To do this we have added a small bank of switches and some code to be able to detect which switches are activated and change the mode. One issue we have been repeatedly encountering is with the programming boxes causing our microcontroller to crash and have issues when plugged into it while had confused us and held up progress on a few occasions. We have made some hardware changes also as the IR was very sensitive and was actually too sensitive for what we needed it to do. Due to this we reverted the resistor values back to the default values instead of the values we had used to get the maximum sensitivity. Finally we also filmed our third video, but ran into a snag with not having the camera pick up any usable audio, so the video had to be filmed again So that’s March gone! Things still moving along well and the code is going well, but its coming down to crunch time so we will have to see if we are still where we need to be next month!

Coding is finally beginning! To try and give myself some help and a head start I have been making some flow diagrams for the code I need to write to hopefully give me some structure and keep me on track. The main code was of course the first step which resulted in a slow start as I was having some trouble getting the ground work set up as I have not yet had much experience coding the micro controller from the ground up but I was helped by the fact we had done some of the micro controller practice exercises previous to this. With a lot of time and help from Augustine we eventually were able to get the main program off the ground! Progress was easier from that point and in a relatively short amount of time we were able to get the code for the bumpers working to a reasonable starting point. From here I moved on to the next part of the obstacle avoidance mode, which is where we had elected to start coding. At this point I have the IR code somewhat working but with some bugs, and it does not function how the group envisioned quite yet so I will continue to work on it. While this work on the code was going on there were some group decisions that as team leader I needed to be a part of, such as the decision of how to mount the shell. We had a couple of different methods to do this, one with a tight fitting shell with bumpers that stick out, and another that used a larger shell that covered the whole mouse, and when the shell was hit would pivot and act as the bumpers. In the end the team elected to go for the first method with the more tight fitting shell due to design and printer constraints, and I was pleased with this decision. While coding we also had some hardware issues, such as some of the ports on the microprocessor lacking solder which was giving us some issues with port connections, but luckily Tom was able to easily fix this. We were also given a battery towards the end of the month meaning that we can now test Clyde without the need for an umbilical power line to the power supply. So progress is really speeding up this month and a lot more options are opening up to us. We have had some issues with the code so far but these seem to be becoming less regular as I become more practiced so hopefully we will have no more hold ups like we did with the main program from now on. Overall I am pleased with progress and hopefully the team and I will keep on moving ahead next month! See you then!

And we return! Finally exams are over! It was a stressful time for sure but now we can get back to robot work! With all of us being away over Christmas and being so busy when we got back with revision and exams no progress of note has been made, other than some ideas that have come out of the inevitable exam procrastination. So whats left of January will be getting ours head back on to the project and trying to move forward. February has a lot of potential and we need to take some big steps to stay on track, so roll on next month!

We finally have some movement! The coding for clyde has begun which means my main role other than that of team manager is finally beginning! So far using some code from our micro mouse interface example that I have modified I have got the clyde off and running! But just in a straight line for now, just as a proof of concept for driving the motors and checking everything works. We are having some trouble with the PWM speed control of individual wheels not working correctly but the code was working for the example so it must be a reasonably simple mistake from the transition. But I am excited with how we are progressing as a group and I am happy to finally be able to code and get working on the brains of our little bot!

This month has been a slower month for me, which has not been the best but my main area of the project other than managing has still not quite yet begun! Although we are edging ever closer. This month I have been doing some smaller jobs and some research tasks. I have designed the bumpers to be attached to the micro switches on the micro mouse, and I am hoping to get them produced this week. I have also been researching filaments that could be used for 3D printing and have the required physical traits for our product, but also be able to defuse the light from the LED’s for us without the need for a secondary diffuser. I have also purchased some RGB LED strips which can be used for testing to see if they are plausible for use in our robot. I also helped out with some documentation task such as recording the results from our IR breadboard testing. So this month has been mainly lots of little jobs, but jobs that needed doing nonetheless! So roll on next month where hopefully I can come into my own.

So overall the first month has gone well, I think I have been reasonably effective in managing the group, I may not have got as much achieved as I would have liked but we are really getting going now so hopefully we will get a lot more done in our next month now we are up and running!