Applications of cognitive robotics in disassembly of products

Abstract

Disassembly automation has encountered difficulties in disassembly process due to the variability in the planning and operation levels that result from uncertainties in quality-quantity of the products returned. Thus, the concept of cognitive robotics is implemented to the vision-based and (semi-) destructive disassembly automation for end-of-life electronic products to handle this variability. The system consists of three operating modules, i.e. cognitive robotic module, vision system module, and disassembly operation module. First, the cognitive robotic module controls the system according to its behaviour influenced by four cognitive functions: reasoning, execution monitoring, learning, and revision. The cognitive robotic agent uses rule-based reasoning to schedule the actions according to the existing knowledge and sensed information from the physical world in regard to the disassembly state. Execution monitoring is used to determine accomplishment of the process. The significant information of the current process is learned and will be implemented in subsequent processes. Learning also occurs in the form of demonstration conducted by the expert user via the graphic user interface to overcome unresolved complex problem. The system is able to learn and revise the knowledge resulting in increased disassembly performance as more processes are performed. Second, the vision system module performs recognition and localisation of the components using common features as detection rules. It also supplies other information regarding the operations. Third, the disassembly operation unit module performs the cutting operations. The physical collision can also be detected and resolved by this module. Consequently, the integrated system is flexible enough to successfully disassemble any models of given product type without specific process plans and parameters being supplied. LCD screens are used as a case-study product in this research.