Vision-based State Estimation and Collision Alerts Generation for Detect-and-Avoid

Abstract

This paper proposed a method to estimate noncooperative intruder's state using Electro-Optical (EO) sensors and to generate conflict alerts to avoid the traffic. In civil area, expensive and noble sensors are not suitable to use for Detect-and-Avoid purpose. Low Cost, Size, Weight, and Power (C-SWaP) sensors as a detection sensors are should be used due to market's require. The demand of Low C-SWaP sensor is growing but the technology is not matured so far. The image-processing based robust aircraft detection algorithm is popular using neural network and the algorithm is applied in this research. The detected aircraft's flight state is estimated by Kalman Filter based on vision information to display collision risk. The concept of Distance at Closet Point of Approach (DCPA) is applied to keep DAA Well Clear (DWC). Overall, this paper shows whole process to implement DAA for non-cooperative aircraft using Low C-SWaP EO sensor. For this research, Conflict Prediction and Display System (CPDS) is applied to generate collision alerts and the CDPS is provided by General Atomics Aeronautical Systems Inc. (GA-ASI).