High-precision Low-SWAP Robust Autonomy & Navigation

Existing and emerging autonomous vehicle systems rely extensively on external aids like the Global Positioning System (GPS) and radio communications. We envision truly autonomous capabilities that utilize on-board sensing and computing capabilities for navigation, autonomy and artificial intelligence tasks. Towards this end, we propose to develop new hardware and software platforms that provide substantially more computing power for visual navigation, autonomy and artificial intelligence at the edge. We propose to develop new algorithms and software for robust visual-inertial navigation that will allow small vehicles to navigate in challenging environments including dust, dirt, large water surfaces, reflective surfaces, moving objects like leaves blown by wind, among many others. We also propose to develop new computing hardware, algorithms and software for high-resolution and high-frame-rate navigation and autonomy, so that the resulting systems achieve operational speeds, e.g., 30 mph in indoor environments and 120 mph in outdoor environments while traveling at 10 feet altitude.á á The key approach is to tightly integrate sensing hardware, computing hardware, algorithms and software together, while we design and build them from the ground up. This tight integration delivers performance that is far beyond what is available when software is designed for and deployed on off-the-shelf computers and sensors. The tight integration entails that we develop custom computing hardware, e.g., efficient circuit boards and effective circuitry that is implemented on Field Programmable Gate Arrays (FPGAs). á