A robot that can drive autonomously, actively seeking more information about the environment as it attempts to infer it, has significant value in many application areas. Range scanners and depth sensors are one of the most popular sensors used in mobile robotics to accomplish several higher level tasks such as local planning, obstacle avoidance, and mapping and localization among others. For any application, it has been observed with laser range-scanners and depth sensors, that the sampling density, i.e., the number of range measurements per unit length of the scanned contour, can vary greatly even within a single scan measurement. The number of samples and their distribution are important factors, for example, when estimating the alignment between two range scans obtained from two different positions. In this paper, an on-line placement algorithm is proposed that computes where the robot must move next so that it is able to sample the environment uniformly and densely. The algorithm guarantees that a minimum number of measurements per unit length of the observed space is obtained, i.e. a high spatial measurement density. At any given time instant the robot computes a NEXT-BEST-VIEW relative to its current position while satisfying a locally-defined constraint function based on the sampling density of points. Two variants of this algorithm, suitable for different practical applications are demonstrated with experiments on real robots in interesting scenarios.