Predicting sUAS conflicts in the national airspace with interacting multiple models and Haversine-based conflict detection system

Abstract

In this paper, a conflict detection system for small Unmanned Aerial Vehicles (sUAS), composed of an interacting multiple model state predictor and a Haversine-distance based conflict detector, is proposed. The conflict detection system was developed and tested via a random recursive simulation in the ROS-Gazebo physics engine environment. The simulation consisted of ten small unmanned aerial vehicles flying along randomly assigned way-point navigation missions within a confined airspace. Way-points are generated from a uniform distribution and then sent to each vehicle. The interacting multiple model state predictor runs on a ground-based system and only has access to current vehicle positional information. It does not have access to the future way-points of individual vehicles. The state predictor is based on Kalman filters that utilize constant velocity, constant acceleration, and constant turn models. It generates near-future position estimates for all vehicles operating within an airspace. These models are probabilistically fused together and projected into the near-future to generate state predictions. These state predictions are then passed to the Haversine distance-based conflict detection algorithm to compare state estimates and identify probable conflicts. The conflicts are detected and flagged based on tunable threshold values which compare distances between predictions for the vehicles operating within the airspace. This paper discusses the development of the random recursive simulation for the ROS-Gazebo framework and the derivation of the interacting multiple model along-with the Haversine-based future conflict detector. The results are presented via simulation to highlight mid-air conflict detection application for sUAS operations in the National Airspace.