Beaconless navigation for mobile robots using grid line pattern

The dead-reckoning of a wheeled vehicle using only an odometer is susceptible to drift errors when no position fixes are provided by external navigation beacon systems. Grid line patterns which can often be found in a structured environment may provide relative pose information of the robot with respect to the grid structure, and this information can be used to correct drift error. Grid line patterns are repetitive in nature, which still leads to the estimation difficulty involving multiple hypotheses. A particle filter was employed and suitable probabilistic sensor models were introduced to handle such multiple hypotheses. Experimental demonstrations were carried out to check the validity of the proposed navigation algorithm.