logo motion mind
Spatial Teaching

When a task is defined for a robot, the task's operation area "A" is also determined as a part of the task definition. The robot must know the geometrical relations of the area to execute the task; for instance, the robot should know where a geometrical element is located, and under what name.

A robot is said to understand an operation area "A" if the robot possesses the following information:

1. The boundary of "A".

2. The position, name, and attributes of each geometrical element in "A"; an element can be a point, a room, a corridor, a doorway, an area patch, or others.

3. The connectivity between each pair of geometrical elements in "A".

How can we endow a robot with this spatial information body? There exists one and only one solution! A human master teaches the robot. Why?

Reason 1: It is impossible for an autonomous robot to unambiguously understand the global geometrical relations of an operation area.

Reason 2: Only the master knows the precise operation area for the given task, and recognizes geometrical elements in the area.

Reason 3: Only the master can give the robot names and other attributes associated with the geometrical elements.

In the teaching phase, the robot is not autonomous, but is obedient to the master.

We cannot rely on robots' poor IQ in their spatial-understanding ability. For instance, no robot can survey and memorize the layout of the first floor of Macy's in Manhattan by itself. For a human, on the other hand, this task is a piece of cake. Therefore this teaching method is very different from the conventional SLAM (Simultaneous Localization And Mapping) approach using an autonomous robot only.

 
© 2010 MotionLab, LLC. Website Design & Web Development: www.businessprodesigns.com