WEBVTT
Kind: captions
Language: en

00:00:00.149 --> 00:00:03.000
A robot now is deployed in a new warehouse.

00:00:03.000 --> 00:00:06.620
How can the robot create a map with the doors
as landmarks?

00:00:06.620 --> 00:00:11.329
Again the robot can move on a line and has
a camera to detect a door.

00:00:11.329 --> 00:00:14.490
In addition, it has its location through GPS.

00:00:14.490 --> 00:00:18.840
We can model the belief on where the new discovered
landmarks are as a probability distribution

00:00:18.840 --> 00:00:21.160
over the possible locations.

00:00:21.160 --> 00:00:24.539
Just deployed, there is no landmark known.

00:00:24.539 --> 00:00:29.011
As the robot gets an image from the camera,
it detects a door and places a new landmark

00:00:29.011 --> 00:00:30.250
in the map.

00:00:30.250 --> 00:00:36.690
Because of uncertainty in the robot's location,
also the landmark has a probability distribution.

00:00:36.690 --> 00:00:41.440
As the robot moves, it discovers new landmarks
which are added in the map with their probability

00:00:41.440 --> 00:00:44.050
distributions, resulting in a final map.