02. Online SLAM

Online SLAM

Online SLAM Problem

  • At time t-1 , the robot will estimate its current pose x t-1 and the map m given its current measurements z t-1 and controls u t-1 .
  • At time t , the robot will estimate its new pose x t and the map m given only its current measurements z t and controls u t .
  • At time t+1 , the robot will estimate its current pose x t+1 and the map m given the measurements z t+1 and controls u t+1 .

This problem can be modeled with the probability equation p(x t , m | z 1:t , u 1:t ) where we solve the posterior represented by the instantaneous pose x t and the map m given the measurements z 1:t and controls u 1:t . Thus, with online SLAM we estimate variables that occur at time t only.

How will the robot solve the Online SLAM problem at t+2 ?

SOLUTION: The robot will estimate the “x t+2 ” pose and map “m” given the measurements “z t+2 ” and controls “u t+2 ”

Next Concept