Mobile target tracking by networked uninhabited autonomous vehicles via hospitability maps

An air vehicle detects a target at some position using its own sensor but delays attack. While the target is detected the air vehicle takes several looks at the target in order to classify it. Once the classification of the target is done the same air vehicle or another vehicle attacks it. During the process of classification the target has moved from its initial location. Since the target has moved away where should the UAVs look for it? This is a prediction and a search problem. Prediction uses the past information and search looks for the target in the predicted locations. As time increases the difference between prediction and search becomes blurrier. Therefore, the research objective now becomes more challenging. We need to explore innovative modeling and estimation techniques that result in more robust estimation and more robustness to model uncertainties. We can rely on terrain based state prediction to determine the likelihood of the new target position, but the probability density function (pdf) of the target position is propagate, as the target moves. The propagation of the pdf for the target position can be non-linear, non-Gaussian, and terrain dependent. The effects of such terrains are captured by something called hospitability maps. A hospitability map provides a likelihood or a "weight" for each point on the terrain's surface for the target to move and maneuver at that location. We introduce the idea of a time varying hospitability map by making the vehicles searching for the targets part of it.