A sliding window approach to exploration for 3D map building using a biologically inspired bridge inspection robot

Abstract

This paper presents a Sliding Window approach to viewpoint selection when exploring an environment using a RGB-D sensor mounted to the end-effector of an inchworm climbing robot for inspecting areas inside steel bridge archways which cannot be easily accessed by workers. The proposed exploration approach uses a kinematic chain robot model and information theory-based next best view calculations to predict poses which are safe and are able to reduce the information remaining in an environment. At each exploration step, a viewpoint is selected by analysing the Pareto efficiency of the predicted information gain and the required movement for a set of candidate poses. In contrast to previous approaches, a sliding window is used to determine candidate poses so as to avoid the costly operation of assessing the set of candidates in its entirety. Experimental results in simulation and on a prototype climbing robot platform show the approach requires fewer gain calculations and less robot movement, and therefore is more efficient than other approaches when exploring a complex 3D steel bridge structure.

Key feature-based approach for efficient exploration of structured environments

“…In the case of a robot inside a tunnel environment it is necessary to detect the surrounding tunnel. Template-based manhole plate or tunnel detection [6] has been shown to robustly detect and generate an environment map based upon prior knowledge of the context of an application, provided that at least two sets of parallel walls can be detected simultaneously.…”

“…Our plane growing algorithm [6], based on [13] and [14], uses an "organised" point cloud with associated normals. Seed points are selected and tested against neighbouring points to try and combine them into a larger plane group and update the plane model.…”

“…Maximal C-space Entropy Reduction (MER) involves finding the Next Best Viewpoint (NBV) that most reduces C-space entropy [3]. Other approaches greedily select a NBV based on estimates of how much information can be seen from a set of candidate viewpoints [4]- [6].…”

“…In this environment a key feature is a view that contains multiple surfaces and surface intersections. The ideal case is to observe all four planes of the tunnel simultaneously [6]. This can be achieved by analysing the partial features in the environment and generating robot motions that allow subsequent observations to improve the view of key features.…”

“…Intuitively, there is a difference when attempting to inspect and map a surrounding environment where the robot and sensor are "inside-lookingout". An alternative strategy is to identify the location of key features in the environment, to which a template can be fitted, encapsulating prior knowledge of the environment [6], [10].…”