AI-based safety at construction sites

There are many work-related accidents in the construction sector.

Statistics from the Swedish Work Environment Authority indicate that there are approximately 3,700 accidents annually that have led to sick leave in recent years in the industry.

Fatal accidents are high in the construction sector, of the 24 fatal accidents in 2020, 10 were in the construction sector. With this project, we want to improve safety at construction sites.

Based on the project partners´ experience, the project will focus on early detection when staff at the construction site are on their way into dangerous situations and reduce the risk through "nudging".To achieve the overall goal, this project focuses on two focus areas introducing AI on construction sites to identify deviant behavior among workers on construction sites and safety-related changes in the environment and situational interaction between the infrastructure / detector system and end users (construction site staff) to improve and maintain security.The project is based on tracking and prediction of road users´ behavior from trajectories for e.g. pedestrians who in turn were used in machine learning-based software to predict road users´ intentions.

From the discovered intentions, the system can understand what is happening in the area.

In addition, it could inform and warn specific people about the deviant and potentially dangerous situation.Data collected (trajectories and image data) can be used to develop and validate a model that can form the basis for creating solutions based on the principles of so-called nudging to contribute to safer behavior among those working on construction sites.The results can be published in magazines such as International Journal of Environmental Research and Public Health and Computers & Industrial Engineering or conferences Intelligent Systems Conference and Artificial Intelligence in Architecture, Engineering and Construction.