[map]POMDP, an online demo: user interface guide

My POMDP pages:

.

User Interface guide

This is a screenshot of the simulation in action. The various components of the interface are described in the table below.

Screenshot of the Flash movie.
ElementDescription
element

Reinitialize the game. The robot is moved back to the start case. Let the game begin !

element

Perform one iteration, which consists in:

  • The robot determines the case where it is most likely to be. That case determines the action the robot is going to take (see the Toggle button).
  • An animation shows the outcome of its action (see the Fast, Normal and Slow buttons). What the wall-sensors detect is also displayed.
  • The robot updates its belief of its position in the labyrinth. These probabilities are translated into gray levels on the map.
element

Perform repeatedly the iteration described above. Loop until the user interrupts it or until the robot reaches the checker.

element

Interrupts the loop started with the Play button.

element

Show or hide some arrows that help visualizing what action the robot is going to take depending of where it thinks to be.

element

Adjust animation speed when showing the outcome of the robot 's action. The animation speed also depends on the speed of your computer.

robot

The robot. Sorry, it 's actually a tank. If you can think of a more peaceful graphism, please let me know.

element

The checker. The robots aims at reaching it.

element

An empty case in the labyrinth. Because it is red, we can say that the robot thinks it has no chance of being at this place.

element

A wall. The robot just can't go there.

element

A few cases. The various gray levels mean various probabilities to be there: the darker, the more likely the robot thinks it is there.

element

The case where the robot thinks it is. Among all the cases where the robot thinks it can be, this one is the most likely. The lighter the yellow, the stronger the belief to be here. When the robot is not on the yellow case, it is "lost". This case determines what action the robot is going to take on the next iteration (see the Toggle button).

element

The wall-sensors. Each sensor is black when a wall is detected and light gray (as in this example) otherwise. In the middle, two indicators shows whether the robot ran into a wall. This indication is given to the user, but the robot doesn't know whether it bumped a wall or not.

Back to top.