Hi! I am a postdoctoral research fellow at Queen's University Belfast, currently a member of the Re-Imagining Engineering Design (RIED) team. The fundamental goal of RIED is to rapidly generate new engineering ideas and concepts using AI and bioinspired methodologies, where both the product and its associated manufacturing system are designed concurrently.
My research interests involve the application of AI for engineering design process, exploring the use of 3D deep learning, computer vision and reinforcement learning. I obtained my PhD degree for work around learning from B-Rep CAD models using graph convolutional neural networks. The majority of my PhD work resulted in the Hierarchical CADNet paper and the MFCAD++ dataset.
Authors: Edgar Buchanan, Simon J. Hickinbotham, Rahul Dubey, Imelda Friel, Andrew R. Colligan, Mark Price & Andrew M. Tyrrell
Conference: IEEE World Congress on Computational Intelligence, 2024
Study of the effects of quality and diversity (QD) algorithms in Evolutionary Development (EvoDevo) processes for engineering design.
Authors: Rahul Dubey, Simon J. Hickinbotham, Andrew R. Colligan, Imelda Friel, Edgar Buchanan, Mark Price & Andrew M. Tyrrell
Journal: Artificial Life, 2024
Exploration of different gene regulatory networks (GRNs) for the optimization of structural engineering designs in an Evolutionary-Development algorithm.
Authors: Edgar Buchanan, Rahul Dubey, Simon J. Hickinbotham, Imelda Friel, Andrew R. Colligan, Mark Price & Andrew M. Tyrrell
Conference: Proceedings of the IEEE Symposium Series on Computational Intelligence, SSCI, 2023
A study into how the initial structure of an EvoDevo process influences the ability to learn development rules.
Authors: Simon J. Hickinbotham, Rahul Dubey, Imelda Friel, Andrew R. Colligan, Mark Price & Andrew M. Tyrrell
Conference: Proceedings of the IEEE Symposium Series on Computational Intelligence, SSCI, 2023
A theory for the use of the von Neumann's Universal Constructor Architecture (UCA) as common language between multidisciplinary applications for evolving engineering systems.
Authors: Zhiwei Zhao, Paul Goodall, Andrew West, Andrew R. Colligan, Imelda Friel, Simon J. Hickinbotham, Mark Price & Yan Jin
Conference: 3rd International Conference on Mechanical, Aerospace and Automotive Engineering (CMAAE 2023), 2023
A bio-inspired evolutionary-development approach for modelling and optimizing buffer allocations in a production line.
Authors: Simon J. Hickinbotham, Rahul Dubey, Imelda Friel, Andrew R. Colligan, Mark Price & Andrew M. Tyrrell
Conference: Proceedings of the IEEE Symposium Series on Computational Intelligence, SSCI, 2022
Applying the biological concept of Evolutionary Developmental (EvoDevo) to evolve growth rules via neural networks to parameter truss structures for multi-objective problems.
Authors: Andrew R. Colligan, Trevor T. Robinson, Declan C. Nolan, Yang Hua & Weijuan Cao
Journal: Computer-Aided Design, 2022
Proposed a graph convolutional neural network architecture for learning from B-Rep CAD models and a complex CAD model dataset called MFCAD++ with labeled machining features.
Authors: Andrew R. Colligan, Trevor T. Robinson, Declan C. Nolan & Yang Hua
Journal: Computer-Aided Design and Applications, 2021
A method of generating labelled point clouds from B-Rep CAD models suitable for machine learning applications.
Authors: Weijuan Cao, Trevor T. Robinson, Yang Hua, Flavien Boussuge, Andrew R. Colligan & Wanbin Pan
Conference: ASME International Design Engineering Technical Conferences & Computers and Information in Engineering Conference, 2020
First approach to propose the use of graph neural networks to learn from B-Rep CAD models. Introduced the benchmark MFCAD dataset for machining feature recognition.
A list of different code projects in which I have worked on or have been involved in.
Code from a MEng Master's project by David Philpott in which I was a supervisor. The repo contains open-source code for using relational learning to cluster together similar entities in CAD models.
Experimental code for generating triangular meshes from micro CT scans. The code was used to generate meshes of titanium lattices from micro CT scan data.
A TensorFlow 2 implementation of the FeatureNet paper. This approach can classify and segment machining features in CAD models by converting the B-Rep to voxel models and using 3D convolutional neural networks.
andrewcolligan.ai@gmail.com
