No GPS, No Problem: Exploiting Signals of Opportunity for Resilient and Accurate Autonomous Navigation in GPS-Denied Environments
Autonomous vehicles rely on a steady stream of signals and information from external sources for localization, route planning, perception, and situational awareness. This includes reliance on positioning, navigation, and timing (PNT) information from global navigation satellite systems (GNSS). Current autonomous vehicles are too trusting of such PNT information and too fragile in the face of loss or attenuation of communications links. There is a global trend of increasing radio frequency interference, whether accidental or deliberate. Civil GNSS signals jamming and spoofing have evolved from a hypothetical threat to an experimentally-verified vulnerability, to an emerging public safety hazard.
This talk will present a framework for resilient and accurate autonomous navigation by exploiting ambient radio frequency signals of opportunity, which are not intended as navigation sources. This framework is termed radio simultaneous localization and mapping (radio SLAM). Here, specialized vehicle-mounted radios collaboratively draw relevant positioning and timing information from ambient signals of opportunity to build and continuously refine a spatiotemporal signal landscape map of the environment within which the vehicles simultaneously localize themselves in space and time. We will present an end-to-end approach, spanning theoretical modeling and analysis of signals of opportunity, specialized software-defined radio (SDR) design, practical navigation algorithm development, and experimental demonstration of radio SLAM on ground vehicles and unmanned aerial vehicles (UAVs), navigating to a meter-level and submeter-level accuracy, respectively. We will also demonstrate the efficacy of radio SLAM in a real-world GPS-jammed environment and on high-altitude aircraft.