Context-aware Deployment of Collaborative Robots

Small and Medium Enterprises (SMEs) produce approximately 70% ofthe world’s products and services. As manufacturing increasinglyshifts towards automation, SMEs must adapt to remain competitive. Collaborative robots (cobots) offer a cost-effective automation solution compared to traditional industrial robots. However, the pre-deployment anddeployment phases of cobot automation are complex and time-consuming,requiring expert knowledge.

This industrial PhD research project aims to simplify cobot automationdeployment by reducing the expertise required. The proposed system design leverages user knowledge of the production process to collect contextaware data of the workcell. To enable the proposed system design, we utilizemobile-based Augmented Reality (AR) 3D sensing for point cloud collectionand custom on-device processing tools, achieving 1 cm accuracy of the captured scans. Moreover, we introduce marking tools to collect context dataof the captured workcell that we use in the deployment phase for optimalcobot base placement, and in the pre-deployment phase for technical feasibility analysis. Thorough system and user studies demonstrate significanttime improvements, with non-expert users completing reachability analysisin under 10 minutes. Additionally, we explore deployment challenges by using Large Language Models to simplify robot programming and developingan AR manual to guide hardwiring the cobot with the automation solution.

The findings show that AR-enabled tools can significantly reduce the complexity and cost of cobot deployment, making automation more accessibleto SMEs. This work contributes to Human-Robot Interaction by providinguser-friendly interfaces that empower non-expert users to effectively deployand manage cobot systems.