Experimental Investigation and Queuing Network (QN) Modeling of Speed-Accuracy Tradeoff (SAT) in Human Prediction of Robot’s Target Selection in Human-Robot Collaboration

Abstract

Human-Robot collaboration (HRC) is playing a pivotal role in modern industry. We conducted human experiments and computational modeling with the Queuing Network (QN) cognitive architecture to investigate the patterns of speed-accuracy tradeoff (SAT) and speed-confidence tradeoff (SCT) in human prediction of a robot’s movement intention. Experimental results show specific patterns of SAT and SCT, which are both affected by task difficulty. For example, clear quantitative relations of SAT are shown (a) in all the easy task conditions, (b) only in the medium to long duration conditions of the medium difficulty situations, but (c) not in any of the hard (most difficult) conditions. To account for SAT and SCT, entity departure processes of the QN are used to represent information accumulation in the human mind, with the entities representing the possible robot movement target locations. This modeling work goes beyond the previous QN models that focused on the arrival and service processes of information entities.