Multi-sensor based intrusion detection for Aerial Work Platforms

Job sites are not always at ground level. Sometimes, workers need to use aerial work platforms, to reach their work. In this case, the risk of falling and the hazard of dropping objects on to coworkers below is concrete. Moreover, pedestrians may also be present in the area where the vehicle is operating, and potentially dangerous interference between pedestrians and the machine operations needs to be considered.

This project was funded by an industrial partner who wanted to create safe environments for workers operating above aerial platforms, and people working below.

The project span over 9 months, and delivered a complete AI-based monitoring and alerting solution to improve the safety of human operators.

The solution comprised a stereo camera paired with custom algorithms for people and object recognition, and a rugged tablet available for the operator. On the specific use case, two systems were integrated into the aerial work platform, one in front and the other one on the back, to monitor a big portion of the area surrounding the vehicle.

Result of a 3D Point Cloud processing during tests in indoor environments

Images acquired by the stereo cameras are fed into the AI-based vision systems, and processed at runtime to produce the 3D point clouds. Thousands of points are then grouped through smart clustering algorithms, and finally, the virtualized objects in space are created. All the objects out of a threshold distance are ignored, since they are far from the working area and are subject to no risk. Objects that are close to the vehicle are instead classified with increasing levels of risk depending on the distance. In this case, both an acoustic and a visual signal are sent to the worker through the graphic interface running on the tablet.

Would you like to know more? Get in touch with us!

Share this post

Share on facebook
Share on twitter
Share on linkedin
Share on print
Share on email