Electric wheelchairs are difficult to maneuver for many disabled people. In a paper presented at the 1st International Workshop on Advances in Service Robotics (March 2003), researchers applied Bayes’s Rule to evaluate an “intelligent” robotic controller that aims to capture the intent of the wheelchair user and aid in navigation. Consider the following scenario. From a certain location in a room, a wheelchair user will either (1)

turn sharply to the left and navigate through a door, (2) proceed straight to the other side of the room, or (3) turn slightly right and stop at a table. Denote these three events as D (for door), S (for straight), and T (for table). Based on previous trips, P(S) = .5, P(S) = .2, and P(T) = .3. The wheelchair is installed with a robot-controlled joystick.

When the user intends to go through the door, he points the joystick straight 30% of the time; when the user intends to go straight, he points the joystick straight 40% of the time;

and when the user intends to go to the table, he points the joystick straight 5% of the time.

If the wheelchair user points the joystick straight, where is his most likely destination?

Bayes’s Rule Application—Wheelchair Control