Estimating observer motion from time-varying range data and fusing this data into a coherent map of the environment are two important problems in robot navigation. Current methods first determine a correspondence between range measurements acquired from different viewpoints, and then compute a motion estimate from this correspondence. In this paper, we present an alternative technique which does not assume that any such correspondence exists. Instead, a smooth surface assumption is used, i.e., the sensed mpoints are assumed to lie on some piecewise smooth surface. A motion estimate is obtained by finding the geometric transformation which makes it most likely (in a Bayesian sense) that the points came from the same surface. We derive an energy equation which measures the distance between the new data points and the dense interpolated depth map which is being incrementally refined. The shape of the energy equation in the neighborhood of the optimal motion estimate is used to compute the uncertainty in the estimate. The resulting motion estimation algorithm can be used in conjunction with other motion estimation systems, and provides a flexible and robust method for computing motion from sparse range data.
In Second International Conference on Computer Vision (ICCV'88)
Publisher IEEE Computer Society Press