Optimal Collision Avoidance Trajectories for Unmanned/Remotely Piloted Aircraft

Abstract

The post-911 environment has punctuated the force-multiplying capabilities that remotely piloted aircraft (RPA) provide combatant commanders at all echelons on the battlefield. Not only have unmanned aircraft systems (UAS) made near-revolutionary impacts on the battlefield, their utility and proliferation in law enforcement, homeland security, humanitarian operations, and commercial applications have likewise increased at a rapid rate. As such, under the Federal Aviation Administration (FAA) Modernization and Reform Act of 2012, the United States Congress tasked the FAA to "provide for the safe integration of civil unmanned aircraft systems into the national airspace system as soon as practicable, but not later than September 30, 2015." However, a necessary entrance criterion to operate RPAs in the national airspace system is the ability to Sense and Avoid (SAA) both cooperative and non-cooperative air traffic and attain the same target level of safety as a traditional manned aircraft platform. The goal of this research effort is twofold: First, determine best techniques for calculating optimal avoidance trajectories, and second, determine best techniques for estimating an intruder aircraft's trajectory in a stochastic environment. This paper describes the optimal control problem associated with SAA and uses a direct orthogonal collocation method to solve this problem and then analyzes these results for deterministic and stochastic collision avoidance scenarios.

Uncertainty Corridors for Three-Dimensional Collision Avoidance

“…This number of particles represents a tradeoff between the desired processing speed and estimation accuracy. This tradeoff, which is processor and application dependent, is further addressed in [27].…”

“…The observation model consists of relative range, range-rate, azimuth, and elevation measurements between the intruder and ownship that are nonlinear combinations of the state estimates. These models are fully described in [27]. Based on these models, the particle filter then generates a point cloud distribution that represents the probability region for the intruder's position at each time step.…”

Collision Avoidance Strategies for Unmanned Aerial Vehicles in Formation Flight

“…However, receding horizon control has some inherent disadvantages coming from absence of state information over the finite time horizon. Smith et al developed a general framework to pose a collision avoidance problem of remotely piloted aircrafts as an optimal control problem using a stochastic estimator, [17], [18] where they used particle filter to minimize the effects of uncertainty caused by pop-up circumstances and to enable real-time implementation. Sujit et al developed a path planning algorithm for multiple UAVs with limited sensor and communication ranges.…”

Particle Filtering for Obstacle Tracking in UAS Sense and Avoid Applications

“…On the other hand, innovative filtering techniques developed for nonlinear systems, such as particle filters, are expected to provide more accurate estimates of obstacle dynamics than EKF, thus improving the accuracy of the DCPA estimate and potentially reducing the delay in collision detection [10]. …”

“…In general, while EKF is a well-assessed technique suitable for real time implementation, nonlinearities in obstacle dynamics and/or in the measurement equation can cause some loss of accuracy in obstacle tracking performance. On the other hand, innovative filtering techniques developed for nonlinear systems, such as particle filters, are expected to provide more accurate estimates of obstacle dynamics than EKF, thus improving the accuracy of the DCPA estimate and potentially reducing the delay in collision detection [ 10 ].…”