Micromouse

I started to take an interest in micromouse after a trip to Tokyo for the 18th All Japan Micromouse Competition, Nov 1997. The Koreans dominated the competition and it was thrilling to watch their mice ran.

My first two mice, Ning and Ning2, are stepper motors based. They are named after my eldest daughter, Ng Ning. My other mice, Min , Min2, Min3, Min3G, Min4 and Min4A are all DC motors based and they are named after my second daughter, Ng Min.

Nov 2001 - Ning2 suprisingly clinched the champion position in the 2001 All-Japan Micromouse Competition. It managed to beat 10 other Korean mice, 1 American mouse and 10 Japanese mice. I'm glad that this old mouse of mine still has what it takes to win.

Nov 2003 - My newer DC mice are fitted with electronic gyroscopes. With the gyro, the turns are much more predictable, and this helps to improve the overall speed .

Nov 2005 - My latest mouse Min4A has a rather disappointing trip to Japan, 2005. Due to a known inherent problem, the mouse started off with a super slow 45 degree right turn. The turn normally takes about 0.1 second, but took 0.6 seconds during the final run. Though I'm aware of the problem, I'm not aware that the turn will become slower as the mouse gets faster. Hence, during the fastest run, the first turn was the slowest. It took me only an hour to correct the problem when I got back to Singapore. Guess I'll have to live with this mistake for a long time. Note that I spent the last 3 weeks before the competition modifying the mouse to prepare for flash photography. And there was not a single flash.

Dec 2006 - It was certainly a lucky win for me this year in the Japan Contest. The Nanyang Polytechnic team had a much faster mouse, but their robots are not so reliable and failed to perform during the final round of the contest. This year, there were noticeably lots of camera flashes and I'm glad my mouse was not affected. In fact, it seemed like none of the robots in the final were affected by the flashes. The bright light during the final seemed to have a much bigger effect on them.

Jan 2008 - Just finished SRG. Second again. I was also second in the 2007 Japan Contest. My latest mouse, Min5 is about 5% faster than Min4A. I achieved that by solely bringing down the robot's c.g. To bring the c.g. lower, I had to reduce the robot's weight so that I can use a smaller motor, the 1717 from Faulhaber. The smaller motors allowed me to cut 2 holes in the PCB so that the motors can be placed lower. The plastic gears for the wheels also help. I had some other ideas to try, but I'm not sure whether to invest further as the Japaneses are changing the micromouse to a smaller format, one that is quarter size. The Japan contest is my main motivation, without which, I find it hard to carry on. I also want to mention that I was invited to compete in Taiwan during October 2007. I won first place. I really enjoyed myself there, as the Taiwaneses are really wonderful hosts.

Nov 2008 - Just back from Japan. I manage to win the contest with Min5. There was no software modification on the robot. I had new motor mounting plates which bring the motors down 1.0mm and the PCB down 0.5mm. I'm not sure if that make any real difference to the performance but I believe it did. I had also changed the 5V regulators and skid pads to smaller versions. The tires (Pit Shimizi PS-300) are old. They are the ones that were given to me by either Idani in 2004 or Nakashima in 2006. Anyway, I had used the same tires for nearly 2 years on Min5 and they are still in very good conditions. I want to mention this as I had noticed that many people use new tires during the competition, which may not be really necessary. A good tires cleaning solution is probably better.

Dec 2009 - Since quite a number of people asked me what happened in Japan, I thought I should write something here. I had both classic mouse and 1/2 size mouse entries. The 1/2 size mouse crashed in the prelim due to a sorting optimization bug. Basically, the optimization routine closes dead ends so that the robot can save time avoiding going into dead ends during quick search. But the 1/2 size mouse goal is no longer in the centre and I basically forget to check if the dead end is also a goal.

As for the classic mouse, it was not able to perform in the final as the sensors were not calibrated to the Japan walls. It was calibrated to the less reflective Singapore plastic walls. In the past, that was not a problem as I have side sensors. Now my robot has only 4 sensors, minus the 2 sides (so as to be the same as 1/2 size mouse). This made the wall calibration more critical. I thought it might be a problem before I go to Japan. But Min6 was able to ran during testing and prelim without calibration, though it wasn't exactly in the centre. The bright light must have made it worse. I suspected that it was the sensor calibration problem only after the third run failed. By then, it was too late to do anything. I paid a price for being overly confident.It would have been a total disaster, if not for the line tracer winning. It's the first time I had not won something in micromouse since I started participating the Japan contest and I must admit I'm quite disappointed with myself. I was confident of winning second place, right after Kato. In the end, Kato's worse nightmare came true. His robot couldn't even find the goal! What a year!

Jan 2010 - Won 1st runner-up for both classic and 1/2 size micromouse in All Japan Contest 2010. I was disappointed not to win something in Robot-tracer due to data overrun. My robot software has a limit of 200 markers and there were around 209 markers (including all junctions). It's sad to lose in such circumstances.

Nov 2011 - Won 1st prize for classic and second prize for 1/2 size micromouse in All Japan Contest 2011. There's a bug in the maze algorithm which I updated 2 years ago to make it more flexible. It underestimates the straight run time by half which forces Ning5 to take a long roundabout. After correcting the 'small' mistake, the robot is able to take the shorter diagonal path. The conseravtive estimated time is 4.996 msec if the shorter path is taken. The same bug is also in Min7. Luckily, Min7 wasn't affected by this mistake. My Robot-tracer seems to suffer from false marker detection. I think the robot is still good enough to win, but need some retuning...

Prizes won :

Ning : SRG 1998, CHAMPION(+ fastest mouse award)

SRG 1999, 1st runner-up

Ning2: SRG 1999, CHAMPION(+ fastest mouse award)

Apec 1999, 1st runner-up

Apec 2000, 1st runner-up

20th All Japan, 1999, 2nd runner-up

Apec 2001, 2nd runner-up

22th All Japan, 2001, CHAMPION

Min : 20th All Japan, 1999, 2nd runner-up (joint winner with Ning2)

Apec 2000, 2nd runner-up

Min2 : 21th All Japan, 2000, 1st runner-up

Apec 2001, 1st runner-up

SRG 2001, CHAMPION(+ fastest mouse award)

Apec 2002, 1st runner-up

SRG 2002, CHAMPION(+ fastest mouse award)

Min3 : 22th All Japan, 2001, 3rd runner-up

Apec 2002, CHAMPION (+ fastest mouse award)

23th All Japan, 2002, 1st runner-up

Apec 2003, CHAMPION (+ fastest mouse award)

SRG 2003, 1st runner-up

Min3G : 24th All Japan, 2003, CHAMPION

Apec 2004, CHAMPION (+ fastest mouse award)

SRG 2004, 2nd runner-up(+ fastest mouse award)

Min4 : 25th All Japan, 2004, CHAMPION

SRG 2005, 2nd runner-up(+ fastest mouse award)

SRG 2006, 1st runner-up

Min4A : 26th All Japan, 2005, 1st runner-up

SRG 2006, 2nd runner-up(+ fastest mouse award)

27th All Japan, 2006, CHAMPION

SRG 2007, 1st runner-up(+ fastest mouse award)

Min5 : 28th All Japan, 2007, 1st runner-up

SRG 2008, 1st runner-up(+ fastest mouse award)

29th All Japan, 2008, CHAMPION

Ning3 : 30th All Japan, 2009, Line Tracer Champion

Min7 : 31st All Japan, 2010, 1st runner-up

SRG 2011, CHAMPION(+ fastest mouse award)

32nd All Japan, 2011, CHAMPION

Ning5 : 31st All Japan, 2010, 1/2 size micromouse, 1st runner-up

32nd All Japan, 2011, 1/2 size micromouse, 1st runner-up