Towards quantifying false alarms for effective human robot interactions.
Mohan Rajesh Elara.
Date of Issue2011
School of Electrical and Electronic Engineering
Human robot teams combining the complementary capabilities of robots and humans towards solving complex tasks are gaining wide spread popularity. Accomplishment of these tasks greatly depends on the quality of interaction between human and the robot thereby requiring models and metrics to evaluate human robot interactions (HRI) in relation to performance. The traditional and most popularly adopted approach to this end has been the neglect tolerance model. The major shortcoming of this traditional model is that it presumes ideal conditions in which an operator switches control between robots sequentially based on an acceptable performance level for each robot whilst ignoring any erroneous interactions. In this thesis, the erroneous interactions that inevitably arise in HRI are identified as false alarm interactions, classified and their effects estimated. More specifically, two significant metrics that quantify the effects of false alarm interactions are defined, viz. false alarm time, and false alarm demand. In addition, the neglect tolerance model is extended to accommodate for the additional demands due to false alarm interactions. Extended neglect tolerance model is further expanded for multi-robot systems taking into account the independent or co-operating natures of robots in the team. Traditional neglect tolerance model forms the basis for fan out metric which is adopted as a general index in predicting the maximum number of robots a single operator can handle simultaneously while maintaining performance at acceptable levels. The fan out metric was redefined to account for additional demands due to the occurrence of false alarm interactions.
DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics