A treat to lunch

On Friday, jk and I went to DSO, just like we always do. Finally, on that day, we were given a specific task to embark on. Creating a robot which follows lines going around picking up colored blocks and sorting them in a gathering area. Or rather, when Dr Ang mentioned the idea of it as an application to saving people in a fire, I prefer expressing it as: sorting humans based on their race so as to prevent unrest.

Then Dr Ang treated us to lunch, and knowing that I was vegetarian, he kindly brought Jk, I, mr chan and the indian guy to a vegetarian restaurant in Orchard called Ling Zhi. The food there wasn't too bad, compared to coffee shop vegetarian food stalls, one of them at bugis I remember distinctly I ate at for a week afterwhich I fell sick literally. Anyway, most of them were praises for the food at that restaurant, I guess they have never really eaten vegetarian food before. I felt it was quite okay, but the mocked dishes were not too tasty. Anyway my father commented that it was not a very nice restaurant. Anyway if anyone thought that the food at lingzhi was great, I wonder how they would find Miao Yi vegetarian (speaking of which my family and I just went today, AGAIN). Read a review which mentioned: '(miao yi) surpasses lingzhi at all angles' and that it was good enough to satisfy vegetarians, non-vegetarians and kids. Not just any vegetarian outlet that takes factory ingredients and deep fried them.

But I still say that lingzhi was good enough, especially the dim sum; carrot cake, spring rolls, siew mai...they were really good. Probably the only reason that it surpasses miao yi in this area was that miao yi didn't serve dim sum.

When we returned to DSO, Jk and I started on the given task. It wasn't as easy as it seemed at first judgement. The main problem lied at programming the robot to identify the different types of junctions; bend, T-junction, cross-junction. We tried arranging the flow of logic with a double light sensor system, but then it wouldn't be able to tell the difference between going off the line by accident and meeting a junction. So we tried three, and other problems arised. Then we even considered rotating light sensors to maximise information captured, but it still had it problems, if we used only three light sensors, no matter how we arranged them.

Then we realised that we can plot our current condition on a line, one extreme being hard coding the movements of the robot, the other extreme being able to cater for every single condition that the robot will meet; like varying line thickness etc... and we went too far to the latter.

So we found that if we fix the line thickness, and calibrate the distance between three light sensors such that on a straight line no two sensors will sense the line at any single time, the problem is a lot easier.