A new approach to sensory integration modeling
Gary George, Frank M. Cardullo, Gary George, Frank M. Cardullo · 1997
This paper describes the general methodology and available advanced techniques that are required to develop a unified model which integrates the responses of transfer functions representing each of the separate sensory modalities. These methods include Kalman filtering, federated Kalman filters, neural networks and various fuzzy techniques to implement the integration process. Much individual sensor modeling has been developed by other researchers over the years and is based on psychophysica l data from that time frame. Some work has been done using a classical Kalman filter to model the sensory integration process for the sensory modalities. A newer approach is federated Kalman filters which has been proposed to deal with a similar problem in avionics where an aircraft flight control computer may receive data from several different types of sensor systems. This approach uses a bank of Kalman filters in parallel.