Research Output

Athos - A Model Driven Approach to Describe and Solve Optimisation Problems

  Implementing solutions for optimisation problems with general purpose high-level programming languages is a time consuming task that can only be carried out by professional software developers who typically are not domain experts. We address this problem by developing the Domain Specific Language Athos that allows declarative specification of Vehicle Routing Problems with Time Windows (VRPTW). The model is input to a generator that creates programs to solve the VRPTW in a multi-agent environment (NetLogo) which is further extended with a Genetic Algorithm optimiser.

We discuss the overall Athos architecture and compare the models with the generated code to demonstrate the benefit for developers by discussing general language related considerations. A case study with a published benchmark gives proof for the practical feasibility of our approach.

Beyond the quality criteria discussed in this paper future work will include extensive field experiments with domain experts applying the language to harden the language and improve usability.

  • Date:

    17 February 2019

  • Publication Status:

    Published

  • DOI:

    10.1145/3300111.3300114

  • Library of Congress:

    QA75 Electronic computers. Computer science

  • Dewey Decimal Classification:

    006.3 Artificial intelligence

  • Funders:

    Edinburgh Napier Funded

Citation

Hoffman, B., Chalmers, K., Urquhart, N., & Guckert, M. (2019). Athos - A Model Driven Approach to Describe and Solve Optimisation Problems. https://doi.org/10.1145/3300111.3300114

Authors

Keywords

Genetic Algorithm, Model Driven Software Development, Multi Agent Systems, Vehicle Routing Problems

Monthly Views:

Available Documents