Date of Original Version
Abstract or Table of Contents
Rich scene understanding from 3-D point clouds is a challenging task that requires contextual reasoning, which is typically computationally expensive. The task is further complicated when we expect the scene analysis algorithm to also efficiently handle data that is continuously streamed from a sensor on a mobile robot. We are forced to make a choice between 1) using a precise representation of the scene at the cost of speed, or 2) making fast, though inaccurate, approximations at the cost of increased misclassifications. In this work, we demonstrate that we can achieve the best of both worlds by using an efficient and simple representation of the scene that also obtains state-of-the-art classification performance. Furthermore, as our efficient representation is naturally suited for streaming data, we demonstrate that we can achieve these high-performance predictions at a rate 3x faster than when using precise representations and without any loss in performance.