SAC is a moveable video camera whose unique software enables it to receive and process input from multiple sensors, then automatically aim itself at a target that...
SAC is a moveable video camera whose unique software enables it to receive and process input from multiple sensors, then automatically aim itself at a target that is the most probable source of that input. SAC's efficient and high-performing algorithm is derived from the study of living organisms.
Developed by researchers at the University of Illinois, this patent-pending technology can pick out and track important items in the environment. A breakthrough computer model of a region of the vertebrate brain and a probabilistic model of multisensor fusion combine to give SAC its unique monitoring and targeting capabilities. The brain's superior colliculus integrates multisensor input and guides orienting movement. As applied to SAC, the patent-pending model is organized as a map and input from sensors are applied to the model in spatial register. Each grid location on the map represents a collicular neuron that receives multisensor input from its own specific location in the environment. Each neuron in the model uses its spatially aligned multisensor input to estimate the probability of a target at the corresponding location in the environment. Then the camera aims at the location that corresponds to the model neuron with the highest target probability. SAC uses multisensor input to compute target probability. This feature derives from application of the Bayes Rule Model of Multisensory Enhancement, a probabilistic model of the multisensory responses of collicular neurons that was developed and previously published by one of the collaborating researchers.