Design and simulation of an interactive human robot disassembly system for end of life phones

Introduction With the growing challenge of e-waste and end-of-life phones, a clear demand is required for efficient and scalable recycling solutions. This study proposes a simulation-based and proof-of-concept framework for a collaborative robotic disassembly system that consists of six specialized stations designed to support the systematic extraction and routing of the main end-of-life phone components. The main contribution is the joint integration of a flexible architecture, more than 20 operational and collaborative constraints, and multimodal voice–gesture interaction for dynamic task reassignment and error recovery. Methods This study integrates human-robot collaboration (HRC) and human-robot interaction (HRI) using voice and gesture commands to enable adaptive task reassignment and real-time error correction with reduced downtime. This framework combines simulation-based evaluation, RoboDK modeling, constraint-driven simulation, and HRC scenarios to assess system responsiveness. Over 30 operational and collaborative constraints were mathematically formulated. Results The Constraint-Driven Python simulation produced an average cumulative processing time of 955.37 s for a modeled batch of 10 phones over 200 iterations, corresponding to an average simulated processing time of 95.54 s per phone, with error rates ranging from 3.3% to 4.1%. The hand-sign recognition module based on YOLOv12n accurately detected gestures with confidence scores in the range of 0.91–0.94, thus improving safety and non-verbal communication in HRC centers. The execution of voice commands for actions such as “start” and “you take” showed inference times of 2.3–2.8 s, thus validating the feasibility of real-time HRI with two methods. With scenarios such as a human abandoning a task, a robot misclassifying a device, or switching tasks based on screen conditions, HRC reduced downtime and improved task continuity by enabling robots to autonomously handle failed or delayed human operations. Discussion A brief discussion highlighted our solution in comparison with rigid systems such as Apple’s. The proposed architecture offers greater flexibility when working on variant inputs, combining robotic precision with human adaptability. The results demonstrate the technical feasibility of combining constraint-driven simulation and multimodal HRI for flexible end-of-life phone disassembly, while full industrial and economic validation remains necessary.