Toy demo of POMDP model for Active Perception
This is a toy demo of POMDP model for Active Perception
Get Started
The following packages will be needed.
numpy
matplotlib
Run
Type in
$ python ./main.py
to run this toy demo
Then you will see the following:
Input the angle and distance of a point or just type ENTER.
If you want a new point, which denotes a new object in the view, to appear, then type in its polar coordinate parameters. Otherwise just press ENTER. For example, if you want a point, the polar coordinates of which are (200 degree, 3 meters), to appear, just type
200 3
then you will see in the upper-left graph, there is a black point appearing.
Everytime you input a point or just press ENTER, the demo will sample a scan angle from the distrition of scan angle. If some points are in the scan range, then the distribution of prediction of these points will update, shown in the upper-right graph. After the update, the distribution of scan angle will also be updated in the lower-left graph to maximize the requirement of detecting as much close points as possible.
Author
Minjun Xu minjunxu@andrew.cmu.edu