Utilizing Cloud Solutions for Object Recognition in the Context of Industrial Robotics Sorting Tasks

The rapid advancements in sectors such as Internet of Things (IoT), Artificial Intelligence (AI), Machine Learning (ML), networking and cloud computing are opening the opportunity for integrating these techniques with robots from the manufacturing field, with the promising perspective of improved performance at reduced costs, inspiring the new field of cloud robotics. In this paper, we propose an architecture for integrating industrial robots with Cloud AI object recognition services. For our proposed solution, we explore the results obtained by simulating a sorting task in RobotStudio, communicating with Microsoft Azure Custom Vision API by sending the web service calls from a computer running a Python script. We present the cycle time added by the communication with the cloud service and test the robustness of the system by comparing the results obtained with four sets of images: unaltered, altered with gaussian noise, barrel distortion and both.