engleski

Human Intention Recognition for Safe Robot Action Planning Using Head Pose

An efficient collaborative work between a person and technical system requires a deeper understanding of the human nature including social, cognitive, emotional or any other relationship that the person could have toward the technical system. Such relationships depend also on the case or specific knowledge about the current task they are performing together. Due to safety reasons, increased variability of new products, flexibility, and demands for defect-resistant production, a contemporary production lack such applications where people and robots operate together using social signals or contextual information. The new paradigms that connect vision of Industry 4.0 with artificial intelligence, robotics, and computer networks are inevitably starting the new era of emerging ubiquitous production cells that will be used in factories of the future. Authors propose an addition to a safety framework using a worker intention recognition with head pose information. As an indicator of intentions of a person in everyday communication, besides the experiences that represent a priori knowledge, humans are relying on social signals.

Human-robot collaboration ; Intention recognition ; Action recognition ; Deep neural networks ; LSTM

nije evidentirano