Error Proofing

Overview: German Aerospace engineers COMPANY D, commissioned the Industreweb team to develop an advanced automated safety system to prevent operators from respiratory damage in a busy paint shop environment. Operators in the spray-booth are required to wear a specific type of safety mask which should be in place prior to entering the work area, to protect the wearer from atomised paint residue and other airborne contaminants. In busy periods, operators could easily enter and accidentally start the machine without having their correct masks in place, leading to potentially lethal consequences. COMPANY D wanted to make sure there was no way the machine would operate unless the operator was wearing the correct PPE.

Technologies used: Python Machine Learning Image Model, YOLO Object detection, Industreweb Data Platform.

Approach: An Industreweb Data Platform (Edge Device) was installed in the control panel for the main paint valve which picks up the power signal from the controller. When the operator enters the paint bay and initiates the start sequence for the sprayer, they must first face a camera (standard webcam) which once initialised by Industreweb, pushes a live video image into the YOLO model, where the algorithm uses a convolutional neural network to detect the presence of the correct facemask. If present, the model confirms the mask is in place via the Industreweb platform which in turn initiates the power cycle for the compressor, allowing the operator to continue with the task. The YOLO Machine Learning model was trained with hundreds of images of the correct facemask in various light conditions, on different people and in different settings to ensure 100% accuracy. Industreweb acts as a bi-directional control system to completely automate the safety sequence.