1
00:00:00,610 --> 00:00:03,640
So we now have a rough idea of how we will perform mapping.

2
00:00:04,060 --> 00:00:09,310
We pointed out that we will focus only on the interest points and use some method to save them in the

3
00:00:09,310 --> 00:00:09,640
graph.

4
00:00:10,150 --> 00:00:15,850
One method of performing topological mapping in a 2D maze was given by the great computer vision teacher.

5
00:00:16,150 --> 00:00:16,690
Dr..

6
00:00:16,710 --> 00:00:17,470
My phone.

7
00:00:18,440 --> 00:00:19,040
Hi, Dr..

8
00:00:19,580 --> 00:00:20,240
Dr. Marx.

9
00:00:20,240 --> 00:00:28,280
One positive trait or two they miss occupants rule wise, from left to right, then top to bottom.

10
00:00:28,670 --> 00:00:34,550
And while replacing any encountered node might be connected with vertical or horizontal neighbors.

11
00:00:35,000 --> 00:00:40,460
This means that at a time one node might be connected to ad mix for neighboring nodes.

12
00:00:41,390 --> 00:00:42,410
This is all fine.

13
00:00:42,590 --> 00:00:45,800
But what would happen if we encountered those diagonal roads?

14
00:00:46,220 --> 00:00:52,010
It will not be connected, even though our robot was perfectly capable of reaching that location.

15
00:00:52,550 --> 00:00:59,570
So we modify the Dr. Mike one pass algorithm to accommodate a diagonal node such as top left and top

16
00:00:59,570 --> 00:00:59,960
right.

17
00:01:00,620 --> 00:01:05,810
This means that one node can now be connected to at max eight neighboring nodes.

18
00:01:06,440 --> 00:01:11,810
This allows us to perform those diagonal movements along with the vertical and horizontal movements.

19
00:01:12,230 --> 00:01:15,020
That's a far more practical solution.