Maciej Trzeciak

I am a 4th-year PhD candidate in Engineering at the University of Cambridge, where I work on 3D computer vision with cameras and lidars. My project is sponsored by BP, GeoSLAM, Laing O'Rourke, Topcon Positioning Group, and Trimble.

Specifically, my research concerns improving accuracy in Simultaneous Localization and Mapping (SLAM) for applications in Architectural, Engineering and Construction (AEC) industries. I have experience in robotic state estimation, optimization in 3D involving point clouds and images, deep learning-based camera-lidar fusion and calibration. I am also a SLAM consultant for both academic and industrial projects and a supervisor of a 3rd-year undergraduate module Statistical Signal Processing.

In the past, I was heavily involved in the development of BIM technology, mainly at the Technical University of Munich. Feel free to get in touch with me using:

Email  /  LinkedIn  /  Github  /  Google Scholar

A patent on real-time error prediction for handheld scanners utilizing pose graphs.

Extension of our ECCV 2022 paper

This dataset allows to measure the error of SLAM algorithms on construction sites.

I predict the spatial error of point clouds produced by lidar-based handheld scanners based on the uncertainty level in the underlying SLAM algorithms.

extnesion of our EC3 2022 paper

I reconstruct an indoor scene with very high resolution using my prototypic mobile scanner.

I compare the precision and density of pointclouds produced by static and mobile laser scanners.

An article by the UK's NRFIS briefly introducing my research project and my contrubtions achieved in my first year at Cambridge.

Before coming to Cambridge, I published the following papers while working at the Technical Univeristy of Munich.

	Towards registration of construction drawings to building information models using knowledge-based extended geometric hashing Maciej Trzeciak; Borrmann, A; 26th International Workshop of the European Group for Intelligent Computing in Engineering, Leuven, Belgium, 2019 [ conference website ] Here I explore the problem of aligning 2D engineering drawings with 3D digital models and propose and validate a method based on computer vision.
	BIM4INFRA Guidelines: Part 9 - Exchange of Digial Models using Industry Foundation Classes (IFC) Borrmann, A; Elixmann, R; Eschenbruch, K; Forster, C; Hecker, D; Hochmuth, M; Klempin, C; Kluge, M; König, M; Liebich, T; Schäferhoff, G; Schmidt, I; Maciej Trzeciak; Tulke, J; Vilgertshofer, S; Wagner, B; mediaTUM, Germany, 2019 [ project's website ] [ mediaTUM ] Governmental guidelines on the theory and practicalisities of exchanging digial 3D models using the IFC standard. The project commissioned by the German Federal Ministry of Transport and Digital Infrastructure.
	Design-to-design exchange of bridge models using IFC standard: a case study Maciej Trzeciak; Borrmann, A; 12th European Conference on Product and Process Modelling, Copenhagen, Denmark, 2018 [ conference website ] This paper presents a case study of exchanging bridge models between two commercial modeling applications using the vendor-neutral format Industry Foundation Classes (IFC).
	Towards registration of construction drawings to building information models Maciej Trzeciak Forum Bauinformatik, Weimar, Germany, 2018 [ conference website ] This is my first paper in which I take a closer look at a general model-to-image registration process, and its applicability to the problem of positioning construction drawings against BIM models. This is the first step towards the consistency verification between these two objects.
	Technical report on model exchange between Revit and Allplan using IFC: a case study for a bridge model Maciej Trzeciak; Borrmann, A; Chair of Computational Modeling and Simulation (CMS), TUM, Germany, 2018 This report presents the process of exchanging 3D digital models between two most common modeling software products in the AEC industry, in particular emphasizing the limitations of the modifiability of geometry.

Source code inspiration from this websie.