Sensors and Methods to Handle UAV

Autonomous vehicles navigating the flight deck of an aircraft carrier must recognize which flight deck director is assigned to them and which signals that director intends for them. They must be able to generate and follow a notional path to an assignedspot based on the director's instructions. But relying on the director's signals alone is not enough. They must be constantly aware of their surroundings as conditions change rapidly and unexpectedly. This is a proposal to develop innovative cognitiveperception and intelligent vehicle control techniques for autonomous flight-deck navigation. Our core innovations are 1) a human gesture recognition algorithm for human hand signal input, 2) a hidden-Markov-model-based command interpretation module, 3) arough path-planning method using dynamic mapping, and 4) a learning method for fine motion control. We will develop the algorithms and supporting software in object-oriented toolkit form. This will support easy modification and reuse. From this work,Energid Technologies plans to develop and sell software toolkits in the areas of collision avoidance, path-planning, and gesture recognition. Though the source code will be provided to the DoD, other customers will purchase the toolkit as softwarelibraries and header files. This toolkit will have wide application. One prominent example is the airline industry. The gesture recognition algorithms can be used as a redundancy check for pilots following director's commands. Other applicationsinclude home, factory, and farm environments. Energid has special interest in agricultural applications, in particular, automated harvesting. We believe the path planning and collision avoidance algorithms developed here can be applied to the guidance ofautonomous harvesting vehicles.