Elsevier

Automation in Construction

Volume 117, September 2020, 103229
Automation in Construction

3D mapping from partial observations: An application to utility mapping

https://doi.org/10.1016/j.autcon.2020.103229Get rights and content
Under a Creative Commons license
open access

Highlights

  • An integer linear programming model is proposed for mapping buried utilities.

  • A method for computing the maximum probability of extending two utility ends is given.

  • Incorporation of side connections, detection free measurements and types.

  • 2D and 3D experiments are presented based on both synthetic and real-world data.

Abstract

Precise mapping of buried utilities is critical to managing massive urban underground infrastructure and preventing utility incidents. Most current research only focuses on generating such maps based on complete information of underground utilities. However, in real-world practice, it is rare that a full picture of buried utilities can be obtained for such mapping. Therefore, this paper explores the problem of generating maps from partial observations of a scene where the actual world is not fully observed. In particular, we focus on the problem of generating 2D/3D maps of buried utilities using a probabilistic based approach. This has the advantage that the method is generic and can be applied to various sources of utility detections, e.g. manhole observations, sensors, and existing records. In this paper, we illustrate our novel methods based on detections from manhole observations and sensor measurements.

This paper makes the following new contributions. It is the first time that partial observations have been used to generate utility maps using optimization based approaches. It is the first time that such a large variety of utilities' properties have been considered, such as location, directions, type and size. Another novel contribution is that different kinds of connections are included to reflect the complex layout and structure of buried utilities. Finally, for the first time to the best of our knowledge, we have integrated utility detection, probability calculation, model formulation and map generation into a single framework.

The proposed framework represents all detections using a common language of probability distributions and then formulates the mapping problem as an Integer Linear Programming (ILP) problem and the final map is generated based on the solution with the highest probability sum. The effectiveness of this system is evaluated on synthetic and real data using appropriate evaluation metrics.

Keywords

Utility mapping
Partial observation
Integer linear programming
3D mapping

Cited by (0)