Collision Avoidance for Construction Equipment

Key Technologies: SLAM, object detection, semantic segmentation, photogrammetry, raspberry pi, QT, open3D, openCV, PCL

The Challenge

Our client wanted to create an automated solution for large construction equipment which would avoid collisions with objects or people. They needed to develop proof of concept for a system which could be adapted to different equipment and was fully configurable so that users could change the zone size, location mapping and the types of alerts.

The Solution

To achieve this we created a distributed sensor system which could process detailed data and streaming these data outputs to a central application. This central application could use stereo-depth data to create 3D maps of objects in the environment and would then use object detectors to find unique objects, compute object distances, determine static and dynamic elements of the environment, and evaluate confidence and “collision” distance.

The Impact

Our client developed a powerful platform which makes construction equipment safer and actively reduces construction-related deaths. The platform has been deployed for multiple types of equipment and sensor configurations, and has helped operators detect compliance violations, monitoring use, and improve safety across multiple metrics.

Ready to Start?