Machine learning based target localization for autonomous unmanned vehicles

This invention helps an autonomous vehicle, like a drone, find a specific object. It starts by taking a wide-view picture and uses a first computer model to check if the object is visible. If not, a second computer model identifies a smaller, more probable area where the object might be, and the vehicle then moves to this new area, specifically adjusting its distance for better visibility, with the second model also helping select the best camera settings for the conditions.