Kilian Speiser
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Kilian Speiser has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Jochen Teizer
Technical University of Denmark
4 shared publications - Kepeng Hong
Department of Civil and Mechanical Engineering, Technical University of Denmark (DTU)
2 shared publications - Carl Schultz
Aarhus University Department of Electrical and Computer Engineering
1 shared publication - Frank Boukamp
School of Property Construction and Project Management, RMIT University, Melbourne, VIC 3000, Australia
1 shared publication Guido Maciocci
1 shared publication- Karsten Winther Johansen
PhD Candidate, Department of Civil and Mechanical Engineering, Technical University of Denmark, Kongens Lyngby, Denmark: )
1 shared publication
Latest Publications
- Open Access
- 14 May 2026
- Published by Elsevier in Advanced Engineering Informatics
- Vol. 74
Agentic system for construction safety risk assessments using large language models and knowledge graphs
- 14 May 2026
- Published by Elsevier in Advanced Engineering Informatics
- Vol. 74
Open AccessAgentic system for construction safety risk assessments using large language models and knowledge graphs
- 1 October 2024
- Published by Emerald Publishing in Proceedings of the Institution of Civil Engineers - Management, Procurement and Law
- Vol. 177 (4) , 173-183
Automatic creation of personalised virtual construction safety training in digital twins
The construction industry faces persistent safety challenges resulting in far too many occupational injuries and fatalities. Even advanced measures like virtual safety training methods still fall short in terms of realism. This paper introduces a pioneering approach that exploits digital twins to create virtual training environments (VTEs) tailored for construction safety. The digital twin for construction safety (DTCS) framework seamlessly integrates building information modelling (BIM) models, schedules, resources and hazard data into a prototype using the game engine Unity. It enables the generation of VTEs with minimal user interaction, empowering safety trainers to define personalised scenarios based on specific project parameters and training requirements tailored to the trainees’ needs. The evaluation of this approach through three distinct real-world projects and a case study with trainees demonstrates its efficacy. The results indicate the ability of the DTCS framework to efficiently create VTEs independently of the project once digital project information is available. Moreover, the framework provides meaningful feedback to the trainees by evaluating the performance using runtime data collection and analysis. The collected data promise use for several applications, but future research must validate their effectiveness. Beyond training, the study highlights the utility of VTEs in improving safety design and hazard mitigation strategies through data-driven insights.
- 1 October 2024
- Published by Emerald Publishing in Proceedings of the Institution of Civil Engineers - Management, Procurement and Law
- Vol. 177 (4) , 173-183
Automatic creation of personalised virtual construction safety training in digital twins
The construction industry faces persistent safety challenges resulting in far too many occupational injuries and fatalities. Even advanced measures like virtual safety training methods still fall short in terms of realism. This paper introduces a pioneering approach that exploits digital twins to create virtual training environments (VTEs) tailored for construction safety. The digital twin for construction safety (DTCS) framework seamlessly integrates building information modelling (BIM) models, schedules, resources and hazard data into a prototype using the game engine Unity. It enables the generation of VTEs with minimal user interaction, empowering safety trainers to define personalised scenarios based on specific project parameters and training requirements tailored to the trainees’ needs. The evaluation of this approach through three distinct real-world projects and a case study with trainees demonstrates its efficacy. The results indicate the ability of the DTCS framework to efficiently create VTEs independently of the project once digital project information is available. Moreover, the framework provides meaningful feedback to the trainees by evaluating the performance using runtime data collection and analysis. The collected data promise use for several applications, but future research must validate their effectiveness. Beyond training, the study highlights the utility of VTEs in improving safety design and hazard mitigation strategies through data-driven insights.
- 19 September 2023
- Published by Springer Nature
- p. 201-212
A Digital Twin Model for Advancing Construction Safety
- 19 September 2023
- Published by Springer Nature
- p. 201-212
A Digital Twin Model for Advancing Construction Safety
- Open Access
- 1 January 2023
- Published by Firenze University Press
- p. 155-166
Enhancing the Realism of Virtual Construction Safety Training: Integration of Real-Time Location Systems for Real-World Hazard Simulations
In numerous studies, virtual training for construction safety has been proposed as a promising approach. However, creating realistic training scenarios requires significant resources, encompassing various elements such as sound, graphics, agent behavior, and realistic hazards. Digital Twins have revolutionized this process, and although so far, on a conceptual level only, significantly reducing the associated workload, it is still not exploiting its full potential. In this work, we propose a novel approach that leverages Real-time Location Systems (RTLS) data to simulate the real-world behavior of construction workers and equipment within Virtual Training Environments (VTEs). We aim to create training scenarios with dynamic real-world instead of hardcoded made-up hazardous events. To achieve this, we propose an extension to our Digital Twin for Construction Safety (DTCS) framework that now integrates (a) trajectory data streams of construction personnel and equipment and (b) technical specifications of the construction site work environment, including location and geometry of terrain and surface objects, to simulate real-world hazards in virtual safety training scenarios. Our further contribution is a case study application to explore the DTCS training capacity. Applying a logical filtering algorithm, we can process the RTLS data and ensure that the movements of the workers and equipment within the virtual environment are as realistic and representative as within the real world. This then enables the creation of realistic hazards that trainees can encounter in the training phase. Preliminary results with trainees suggest that the proposed work can have a high potential to enhance the realism of safety training, especially when they need to experience human-machine-related interactions safely. However, further work is required to create more responsive learning environments where the equipment follows real trajectories but also responds intelligently to the trainees' actions. By leveraging real-time data and advanced visualization technologies, we bridge the gap between the physical and virtual realms, enabling trainees to interact and navigate within a realistic virtual environment
- 1 January 2023
- Published by Firenze University Press
- p. 155-166
Open AccessEnhancing the Realism of Virtual Construction Safety Training: Integration of Real-Time Location Systems for Real-World Hazard Simulations
In numerous studies, virtual training for construction safety has been proposed as a promising approach. However, creating realistic training scenarios requires significant resources, encompassing various elements such as sound, graphics, agent behavior, and realistic hazards. Digital Twins have revolutionized this process, and although so far, on a conceptual level only, significantly reducing the associated workload, it is still not exploiting its full potential. In this work, we propose a novel approach that leverages Real-time Location Systems (RTLS) data to simulate the real-world behavior of construction workers and equipment within Virtual Training Environments (VTEs). We aim to create training scenarios with dynamic real-world instead of hardcoded made-up hazardous events. To achieve this, we propose an extension to our Digital Twin for Construction Safety (DTCS) framework that now integrates (a) trajectory data streams of construction personnel and equipment and (b) technical specifications of the construction site work environment, including location and geometry of terrain and surface objects, to simulate real-world hazards in virtual safety training scenarios. Our further contribution is a case study application to explore the DTCS training capacity. Applying a logical filtering algorithm, we can process the RTLS data and ensure that the movements of the workers and equipment within the virtual environment are as realistic and representative as within the real world. This then enables the creation of realistic hazards that trainees can encounter in the training phase. Preliminary results with trainees suggest that the proposed work can have a high potential to enhance the realism of safety training, especially when they need to experience human-machine-related interactions safely. However, further work is required to create more responsive learning environments where the equipment follows real trajectories but also responds intelligently to the trainees' actions. By leveraging real-time data and advanced visualization technologies, we bridge the gap between the physical and virtual realms, enabling trainees to interact and navigate within a realistic virtual environment