Intelligent sensory decision-making for error identification in autonomous robotic systems

Successful automatic assembly of complex artefacts requires the robotic system to have the capability of detecting, identifying and recovering from various errors. Efficient error identification process is essential to ensure fast recovery and minimum loss of production time. It is not cost-effective to interrogate every sensor for every pass through the assembly process. This paper presents a machine-learning approach to identify error. The basic idea is to construct a decision tree based on some sensor and error attributes in the knowledge base.