Previous Abstract Return to Session D4

Session D4: PNT Situational Awareness: Operations

Autonomous Vehicle Path Planning Under Uncertainty
Madison Gillan, Timothy Machin, Daniel Broyles, Air Force Institute of Technology
Location: Room 6-8
Alternate Number 1

This research addresses crucial challenges in contested airspace, focusing on Autonomous Vehicle (AV) flight path planning for Air Force missions. Unlike previous work, which has focused primarily on optimal trajectories with known Weapon Engagement Zone (WEZ) parameters modeled as hard constraints, this research enables strategies for trajectory generation when completely avoiding the WEZ is unfeasible or when WEZ parameters are unknown. Utilizing chance-constrained optimization, the primary objective of this research is to enhance decision-making capabilities by incorporating considerations of uncertainty and risk. The AV initiates its path with a rapidly exploring random sampling algorithm from its start to the goal location. Leveraging stochastic estimation and control, the AV refines its navigation by continuously updating its estimated state based on sensor inputs, allowing for real-time adjustments to uncertainties in the current state location, weapon location, and weapon engagement zone (WEZ) range. During navigation, when encountering potential threats such as enemy weapons, the AV strategically maneuvers within a specified range of the weapon to minimize the risk of capture. The formulation of metrics for both risk and performance considers uncertainties in the current state location, weapon location, and WEZ range, therefore enabling the creation of more resilient control strategies and the ability to generate paths at different assumed risk levels. This research extends beyond deterministic models, allowing for AV navigation in the presence of uncertain threats and ensuring adaptability when handling evolving environments.
A key innovation lies in the use of chance-constrained optimization for flight path algorithms in the context of uncertain weapon engagement zones, which is an essential part of autonomous flight and a large area of interest for the Air Force. Furthermore, this research lays the foundations for future expansion including path-replanning mechanisms to dynamically adapt to changing environments and new or shifting weapon locations. These unique aspects are crucial for advancing the Air Force’s mission to effectively employ autonomous vehicles in uncertain, contested environments. The goal outcome of this research is to enhance the Air Force and DoD’s understanding of how autonomous vehicles can safely navigate dynamic threat environments while optimizing their flight paths for improved efficiency and risk mitigation. In essence, the research findings can potentially empower the DoD to strategically employ autonomous vehicles, ensuring they navigate with precision, efficiency, and heightened risk awareness in complex and dynamic scenarios.



Previous Abstract Return to Session D4