Fast and Agile Perception-enabled Precision Navigation and Autonomy

We envision Unmanned Aerial Systems (UAS) that can navigate contested environments without any external aids, such as GPS, using only on-board sensing and computing, so that they can be completely independent of a base station when such independence becomes necessary. Even within these constraints, we envision UAS that can conduct precision navigation and autonomy as we describe below.  To handle no-external aids, we envision that the new class of UAS can be provided with certain maps or natural language descriptions, if available, and can still handle the navigation and autonomy required to complete the mission. The maps can include satellite imagery or even a hand-drawn conceptual map. Even in the absence of any other tangible data, they can be provided with a direction to go towards and a rough distance to travel (potentially also the description of a target), so that they can, for instance, gather intelligence (within specified objectives and constraints) and come back to their starting location.  To handle contested environments, we envision that they reach and surpass operational speeds, able to navigate at low-altitudes, amongst obstacles (some of which may be moving). We envision them to be relatively very low cost, so that they can be deployed in large numbers and they can be attritable.  To handle independence (when necessary), we envision that they are able to run processing for navigation and autonomy on computers that are carried on board the vehicle using sensors that are entirely carried on board the vehicle. We envision that these vehicles are still integrated, that is, they are ultimately directly controlled by human operators, when human operators can communicate with the vehicle, and in the absence of any communication they still operate within the objectives and constraints set by the human operators prior to any communication cutoff.  Within all these operational objectives and constraints, we envision these UAS to still have precision navigation capabilities, such as the ability to pinpoint a target, identify and track moving targets with precision. Tycho AI team has developed advanced electronics (carrying high-rate optical sensors, high-precision inertial measurement units, and GPU- and FPGA-enabed computers) and advanced algorithms and software (including, optical-inertial navigation, obstacle detection, motion generation, and sophisticated nonlinear control systems) that will form the basis of this project.