Map building using occupancy grids with differentiated occupancy states

Abstract

For map building purposes it is necessary to distinguish between moving and not moving objects, because moving objects might falsify map matching procedures. This paper presents a new approach of a grid based map building algorithm which represents an extension of occupancy grids. As well as it its archetype, the algorithm can easily be used for 2d or 3d time-dependent map building procedures. The original occupancy grids were extended so that cells which are marked as occupied get further differentiated regarding the motion characteristic of their particular obstacle. An occupied cell can therefore be differentiated in a cell occupied by a dynamic, quasi static or static obstacle. This differentiation might also come in handy regarding object detection or similar algorithms. The presented algorithm was developed and tested using a SICK LMS511 high resolution laser measurement system.