Skip to content

Research Opportunity:
Navigation and Estimation

General Category: Navigation and Estimation

Advisor: Kevin Brink, Technical Lead, Navigation and Estimation Section, AFRL Munitions Directorate

Robust GPS-denied Single and Multi-agent (Cooperative) Navigation

  • Signal processing for single-agent navigation; either pre-processing (EO/IR vision, RF IQ data, etc.) or navigation filter measurement processor development (magnetic, other tightly-couple concepts, etc.) which provide position, velocity, attitude, or other relevant aiding to an inertial navigation system when GPS is not available. 
  • ML/AI augmentation of navigation preprocessors and/or filters. Apply ML/AI to replace or augment classic signal preprocessors or underlying filter/navigator structure for improved system performance (i.e., smaller navigation errors, while still providing a valid covariance). Examples include AI/ML for vision aiding or augmenting or even replacing the propagation and measurement models to account for higher order terms lacking in standard Bayesian filter implementations.
  • Multi-agent collaborative navigation concepts with an emphasis of scalability (implying limited communication bandwidth requirements), robustness (to short-term and long-term communication losses, or agent attrition), and limited a priori data requirements (the system should be able to initialize or reinitialize without assuming precise knowledge of initial conditions). Systems should provide accurate relative positioning information of cooperating agents, “disseminate” global position data when any agent has a relevant measurement (e.g., GPS, a vision-based position measurement, etc.) and ideal concepts should gracefully degrade to a state of the art single-agent navigation performance if communication is unavailable and revert back to improved collaborative navigation if communications are restored.
Register for Fair

Questions? Contact:  techtransfer@doolittleinstitute.org