How can we improve safety in the construction industry? In a unique cooperation between NCC, Microsoft, Sigma, Arrow and Edins Byggkranar we have developed digital safety barriers to identify risks when working with heavy crane lifting. In this test we wanted to show that image analysis can be used to interpret what is under and near the crane.
Most large construction sites have a crane that is important part of the workload. Lifting can be difficult with limited visibility, which makes heavy crane lifting a large safety hazard. For the last 10 years, looking at the Nordic countries, the construction industry is one of the industries with most fatal accidents, where one of three high-risk areas are heavy crane lifting. NCC wanted to explore if digital safety barriers could increase safety on construction sites – using AI technology in the cranes.
At the test site, Ångström Laboratory outside Uppsala, cameras, computers and network has been installed on a crane. Cameras attached under the tower crane collects images that is being processed in the computer located in the crane cabin. The Azure based solution has been trained to detect people and vehicles under the crane using deep learning on real time video. The processing is done in the crane cabin to instantly alert the crane operator and the personnel on the ground if they are inside a defined risk area.
By using modern cloud technologies such as AI and Machine Learning we are able to bring new ideas to life. The first step in this test has been to train the AI -model in Azure with images that’s been collected. A simple view has been built to visualize the current situation analyzed by the AI-model. This test has shown that it is possible to detect, and in the end also warn, employees who are under suspended loads or are on their way into a risk zone. It also shows that computer vision works in near real-time, with image processing executed directly in the crane.
“Sigma’s experience and competence has complemented us well in this project. Contributing with key technical skills and advising us how to set up a solution for a valuable proof of concept that also has the prerequisites for a potential production solution“