Complex & Contested Environments
Perceptual programs appear to concentrate on increasing navigational autonomy for individual or related swarms of platforms. Perception and situational awareness to operate in a complex battle space for vehicle missions appears to be a secondary priority.
However, most unmanned combat vehicles are required to operate in environments that are both unstructured and dynamic, where existing maps provide little guidance and both cooperative and hostile agents abound. Developing learning methods that can cope with such complex environments is an important challenge for future R&D.
Here are some challenges that are relevant to operating in complex and contested environments.
- Recent past history in contingency operations has shown that Integrating the perceptions of the individual platforms for understanding the battlespace was not in evidence beyond representing positions with map iconography. Perception to support human comprehension of the platform state and to project its relation to the battlespace and mission objectives is largely ignored and instead erroneously treated as a computer display problem; however, a display cannot compensate for the lack of sensing.
- Airspace deconfliction for dense manned-unmanned system operations – Although sense-and-avoid technologies are vastly improving, they have to be examined in the context of human machine complex relationships. Significant gaps appears to be less in the fundamental theory but rather in hardening these technical and sociomechanical solutions and integrating them with existing technologies and within socioorganizational constraints.
- Real-time pop-up threat detection and identification. Threat detection and identification can be viewed as the highest level of situational awareness, in which the warfighter and swarms of machines can identify and project needed action. Threats can be either detected by onboard perceptual systems for an individual platform or by integration with observations from multiple platforms and information from other sources.
- Multi-sensor integration in complex settings -- Perception for unmanned systems generally relies on a single sensor per capability, for example a range sensor for autonomous navigation and a camera for mission payloads. Multi-sensor integration, either for increased sensing certainty or more comprehensive world modeling, may appear to require further study.
- Today’s systems fly blind with respect to other autonomous air platforms in their vicinity. Just flooding the airspace with simple UAVs flying random patterns
The PowerPoint, podcast and video on this page address the considerations to ensure mission continuation in environments with kinetic and non-kinetic threats.
Autonomy Strategy Slide Deck - Part 4