Tracking Construction Crane
Updated: Nov 4, 2019
Construction crane is extremely common in worldwider construction site. Many construction projects involve the coordination and cooperation of multiple entities to achieve the final as-built structure. Due to the construction process requires all the materials and instrument keeping changing the position, the installment and materials uploading need to be inspected regularly. It is highly related to the construction productivity, safety and cost. All these three issues are also important to be carefully measured.
According to the data, the world crane safety statistics show that the 38% accidents happened during in-operation, 31% accidents are related to climbing, assembly and disassembly, the rest 31% are related to disaster, wind and natural causes.
Therefore, the problems are: how we setup the measurement and control the risk of crane in order to reduce the risk and keep the productivity in the meantime. There are already companies working on the solution. Customindz Viact.ai certainly is one of them, current problem scenarios are helping construction contractors to place IP camera and sensors to certain point to surveillance the crane. Two install-based devices are needed: camera and sensor. In this blog, we will highly focus on the camera surveillance methods and capabilities.
The position of IP camera. Generally there are two solutions, one is to setup the camera to a peak that can overlook the whole crane in a certain distance. Another solution is to install camera on the load or the crane jib/bucket. For example, HoistCam, is the another solution provider, which strongly focus on camera deployment for construction cranes. They designed a battery detached camera and connect it to crane hoist by magnets, therefore, the camera could have a special angle to surveillance the load the frequency.
The capabilities. Video analysis, for example, what viact AI did is to track the activities of crane in a long period and calculate the angle of movement, matching the planned activities recorded from other data sources to infer the activities in the future. Other video analysis also does detect the load type, to infer the activity. However, the distance and the video quality largely impact on the object detection performance. Using the crane jib to infer the activity, therefore, is a good solution.
The methods. Basic background deletion on video/image is necessary, video detection on the boundary of jib is also easy implemented.