Automated Drawing of Railway Schematics Using Numerical Optimization in SAT
Paper in proceeding, 2019

Schematic drawings showing railway tracks and equipment are commonly used to visualize railway operations and to communicate system specifications and construction blueprints. Recent advances in on-line collaboration and modeling tools have raised the expectations for quickly making changes to models, resulting in frequent changes to layouts, text, and/or symbols in schematic drawings. Automating the creation of high-quality schematic views from geographical and topological models can help engineers produce and update drawings efficiently. This paper describes three methods for automatically producing schematic railway drawings with increasing level of quality and control over the result. The final method, implemented in the tool that we present, can use any combination of the following optimization criteria, which have different priorities in different use cases: width and height of the drawing, the diagonal line lengths, and the number of bends. We show how to encode schematic railway drawings as an optimization problem over Boolean and numerical domains, using combinations of unary number encoding, lazy difference constraints, and numerical optimization into an incremental SAT formulation. We compare resulting drawings from each of the three approaches, applied to models of real-world engineering projects and existing infrastructure. We also show how to add symbols and labels to the track plan, which is important for the usefulness of the final outputs. Since the proposed tool is customizable and efficiently produces high-quality drawings from railML 2.x models, it can be used (as it is or extended) both as an integrated module in an industrial design tool like RailCOMPLETE, or by researchers for visualization purposes.


Bjørnar Luteberget

Railcomplete AS

Koen Claessen

Chalmers, Computer Science and Engineering (Chalmers), Functional Programming

Christian Johansen

University of Oslo

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

03029743 (ISSN) 16113349 (eISSN)

Vol. 11918 341-359

15th International Conference, IFM 2019
Bergen, Norway,

Subject Categories

Other Computer and Information Science

Control Engineering

Computer Science



More information

Latest update