research/658-chemotaxis-controller

[jharwell]

Attraction to nearby objects of interest without memory; controllers using it
would be between CRW and DPO in complexity. Controllers using this force to find
objects would be more amenable to mathematical analysis than DPO I think, and it
would be interesting to see how close controllers using this approach could come
to DPO.

The nature of the signal (does it fall off with distance, change over time, etc)
detected and whether or not all signals are averaged together to compute the
chemotaxis force vector would be things to explore.

Blocks/caches/etc could have to give off unique (for the object type) LED
signals which robots would track.

Robots switching between chemotaxis and CRW/levy walk would be able to be fully
modelled mathematically (see Nurzamann2008 to start).