A Hybrid Rough Aggregation Approach for the Selection of Artificial Intelligence-Based Industrial Cleaning Robots Used in Public Spaces From the Perspective of Urban Waste Management

Abstract

Waste management is becoming increasingly complex and challenging, especially in megacities with large populations. Unlike the past, when urban waste was simply collected and disposed of, modern waste management requires careful planning and execution of collection, separation, recycling, and reuse processes. Effective management of this complex system now needs more than just human effort. Integrating artificial intelligence (AI)-based systems into waste management can enhance waste reduction, reuse, and recycling effectiveness and efficiency. Selecting suitable AI-based cleaning robots (AI-ICR) for crowded public spaces, such as stations, train stations, and airports, poses complex decision-making challenges. The primary challenge is the novelty of the technology, which leads to uncertainties in selecting AI-ICRs. To address this challenge, we have developed a decision-making approach based on rough Archimedean-Dombi partitioned aggregation. This approach, termed "rough Archimedean-Dombi partitioned aggregation," combines the flexibility of Archimedean operators, the smoothness of Dombi operators, and the structured decomposition of Partitioned operators. This model is mainly chosen for its ability to handle the uncertainty and complexity inherent in multiple criteria decision-making (MCDM) processes. Leveraging rough numbers provides a robust framework for evaluating AI-ICRs under uncertain conditions. The main advantage of this model is its robustness, consistency, stability, and ability to handle complex uncertainties. We applied the proposed model to assess four AI-ICR alternatives identified through extensive research. We evaluated these alternatives using eighteen criteria established through comprehensive field studies. Based on the results, "Recycling cost (B12)" emerged as the most crucial criterion for selecting AIICRs. Additionally, the research identifies the SD45 manufactured by Peppermint Robotics Co. as the optimal AIICR candidate. Finally, the sensitivity and benchmark analyses to validate the proposed model confirm its robustness, consistency, and reliability.