Intermittent Adaptive Spatial Field Estimation and Concurrent Evacuation Planning Using Field-Dependent Evacuee Guidance

This work combines a path planning scheme used for the evacuation of humans in indoor environments with the real-time estimation of spatially varying fields using adaptive methods. When a hazardous environment is known, then one possible trajectory planning scheme uses level-set methods to guide a human to safety (escape exit) while at the same time minimizes the accumulated amount of the hazardous field modelled as the hazardous substance inhaled. When the field representing the spatial distribution of the hazardous substance in unknown, then an arrested adaptive estimate of the spatial field is proposed in the level-set guidance. The human evacuee viewed as a mobile agent, obtains spatial field measurements and process them in an adaptive learning scheme to obtain an estimate of the spatial field. When a planning period is added to the traveling period, the adaptive scheme obtains the most recent spatial field estimate (arrested adaptation) and uses it as a time-invariant spatial field for trajectory planning.