This paper deals with map-based localization problem. Generally, errors between an environment and a prior-map affects to the localization result. We prepare several types of map with different detailed level. Then we investigate the localization accuracy of virtual sensing image in each map type. We confirmed that the map with lower detailed level is, the worse the localization result is.

Recently we proposed the stochastic sampling method (SSM), which can numerically generate sample points on complicated implicit surfaces quickly and uniformly. In this paper we generalize the method in two aspects: (1) We introduce two kinds of boundary conditions, so that we can sample a finite part of an open surface spreading infinitely. (2) We generalize the stochastic differential equation used in the SSM, so that its solutions can satisfy plural constraint conditions simultaneously. The first generalization enables us to visualize cut views of open surfaces. The second generalization enables us to visualize intersections of static and moving implicit surfaces, which leads to detailed investigation of intersections and other interesting applications such as visualization of contour maps.