Research Article

Bootstrap Learning and Visual Processing Management on Mobile Robots

Algorithm 1

Illumination adaptation algorithm.
Require: For each known illumination ,
 color map , distributions , and distribution
 of JS-distances .
Require: Algorithm to plan motion and learn colors au-
  tonomously (Algorithm 2).
Require: Positions, shapes and color labels of the objects of
  interest in the robot's environment. Initial robot pose.
( )Initialize: , , (no prior
illumination knowledge).
( )Plan motion and learn .
( )Generate , space distributions, and
distribution of JS-distances, , using images cap-
tured at random during color learning.
( )Save image statistics, .
( )while true do
( ) Get new image. Segment image and detect objects.
( ) if minorChange ( ) then
( )  minorUpdate ( ). Get from current color
    distributions.
( )  Revise current illumination representation to get
    and , to be used for subsequent opera-
    tions.
( ) end  if
( ) if   then
( )   distribution of current image.
( )  for     to     do
( )   
( )  end  for
( )  if   Exists ( )  then
( )   
( )  end  if
( )   if   Exists ( ) and withinRange ( , )  then
( )    Continue with .
( )   else  if  withinRange ( , )  then
( )    Continue with .
( )   else  if  withinRange( , ),   then
( )    Use , .
( )   else
( )    New illumination, , .
( )    Learn autonomously.
( )    Learn for new illumination.
( )    Use for subsequent operations.
( )   end  if
( )    .
( )  end  if
( ) end  while