Associative methods for topological path planning

Associative methods for topological path planning 20220209184809 rely on the assumption that a location or landmark has:

• Perceptual stability: views of the location from nearby places should look similar.
• Perceptual distinguishability: views far away should look different.

We can create a behavior that converts sensor observation into the direction to reach a particular landmark. This shares the idea of neighborhood of relational methods 20220221202140 but relies heavily on computer vision.

Example of techniques

Visual homing: partition images on sections, and for each section examine and measure some attributes. Patterns in the neighborhood will be similar, and we can use this patterns to either turn right or left.

Qual nav: localize the robot to a particular orientation of a given pair of landmarks boundaries. Instead of localizing to a certain landmark, localizing to a pair will always provide the same orientation on a certain neighborhood.

Advantages and disadvantages

• ADV: tight coupling between sensing to homing.
• DISADV: more power and memory required.
• DISADV: brittle to changes on the environment of landmarks.

Notes References

20220209184809 Topological path planning

20210514183815 INDEX - Robotics

References

(Murphy 2000)

Murphy, Robin. 2000. Introduction to AI Robotics. Intelligent Robotics and Autonomous Agents. Cambridge, Mass: MIT Press.