 |
Web Hosting by Netfirms | Free Domain Names by Netfirms |
|
|
|
LandMark Navigation
Different alogorithms, such as dead reckoning and wall following, were considered, what was decided to do was to write a navigation algorithm that achieved the best results of both dead reckoning and wall following. Our system, called LandMark Navigation, achieves this by combining the best of both algorithms. A command is issued to the robot to do something, while looking for a certain sensor pattern, while maintaining another sensor pattern. The three parts of this command are the Action, Event, and Requirement. An example would be:
| Action | Proceed forward |
| Event | Until the front sensor detects a wall |
| Requirement | Maintain a constant distance from the left wall |
A simple table is included of the basic Actions, Events, and Requirements, refer to Table 1. Using these simple primitives, a route is programmed through the maze.
Table – Basic Actions, Events, and Requirements
| Actions | Events | Requirements |
| Proceed
forward Turn 180 Turn 90 left Turn 90 right |
Until the X
sensor detects a wall Until the X sensor doesn’t detect a wall Until a timer expires |
Maintain
distance from sensor X No requirement |
The advantage of this method is that no set distances are measured, so that if the maze changes slightly, the algorithm will adapt. The other advantage is that since an optimal route is already provided it will be able to check every room without wasting time. The primary disadvantage is that the general characteristics of the maze need to be known ahead of time. For our purpose this is not much of a drawback since the floor plan is known ahead of time.
Previous: Parts
Next: Specifications
Last Modified: Feb 14, 2000
Comments and Suggestions are welcome!