Anna Elena Tirri

Morphological filtering and target tracking for vision-based UAS sense and avoid

This paper presents a customized detection and tracking algorithm for vision-based non cooperative UAS sense and avoid. Obstacle detection and tentative tracking for track confirmation are based on top-hat and bottom-hat morphological filtering, local image analysis for a limited set of regions of interest, and a multi-frame processing in stabilized coordinates. Once firm tracking is achieved, template matching and state estimation based on Kalman filtering are used to track the intruder aircraft and estimate its angular position and velocity. The developed technique has been tested using flight data gathered in a sense and avoid research project carried out by the Italian Aerospace Research Center and the Department of Industrial Engineering of the university of Naples “Federico II”. Performance evaluated in two near collision geometries allows estimating algorithm robustness in terms of sensitivity on weather and illumination conditions, detection range and false alarm rate, and overall tracking accuracy.

Feature article: Experimental analysis of onboard non-cooperative sense and avoid solutions based on radar, optical sensors, and data fusion

This article summarized processing approaches and presented an experimental analysis of the levels of situational awareness relevant to different sensing architectures for non-cooperative sense and avoid, based on standalone radar, standalone EO, and radar/EO data fusion, respectively. In summary, presented experimental results leave the door open for various sensing solutions. This depends on the different weight, size, power, and cost budgets available for different UAS classes, on the possible approaches for sensing system design (development of ad hoc sensors vs. integration of existing ones), and on the fact that the impact of the different situational awareness levels on collision avoidance safety strongly depends on the unmanned aircraft maneuverability and on the dynamics of obstacles present in the considered airspace scenario.

Sky Region Obstacle Detection and Tracking for Vision-Based UAS Sense and Avoid

A customized detection and tracking algorithm for vision-based non cooperative UAS sense and avoid aimed at obstacles approaching from above the horizon is presented in this paper. The proposed approach comprises two main steps. Specifically, the first processing step is relevant to obstacle detection and tentative tracking for track confirmation and is based on top-hat and bottom-hat morphological filtering, local image analysis for a limited set of regions of interest, and multi-frame processing in stabilized coordinates. Once firm tracking is achieved, template matching and state estimation based on Kalman filtering are used to track the intruder aircraft and estimate its angular position and velocity. An extensive experimental analysis is presented which is based on a large set of flight data gathered in realistic near collision scenarios, in different operating conditions in terms of weather and illumination, and adopting different navigation units onboard the ownship. In particular, the focus is set on flight segments at a range between 3 km and 1.3 km, since the major interest is in understanding algorithm potential for relatively large time to collision. System performance is evaluated in terms of declaration range, probability of correct declaration, average number of false positives, tracking accuracy (angles and angular rates in a stabilized North-East-Down reference frame) and robustness with respect to track loss phenomena. Promising results are achieved regarding the trade-off between declaration range and false alarm probability, while the onboard navigation unit is found to heavily impact tracking accuracy.

Flight Performance Assessment of Vision-based Detection and Tracking for UAS Sense and Avoid

This paper focuses on a vision-based detection and tracking algorithm for UAS non cooperative collision avoidance. It is based on top-hat and bottom-hat morphological filtering, local image analysis for a limited set of regions of interest, and a multi-frame tracking algorithm in stabilized coordinates that is used to confirm intruder detection. Once this detection is achieved, template matching-based algorithms are used to track the intruder in subsequent frames. The developed technique has been tested using flight data gathered in the framework of the sense and avoid research project carried out by the Italian Aerospace Research Center and the Department of Industrial Engineering of the university of Naples “Federico II”. First results are promising: considering two frontal encounters in completely different illumination conditions, range for confirmed detection of an ultra-light intruder aircraft is in both cases of the order of 2.5 km, no false tracks are generated, and angular measurement accuracy in North-East-Down coordinates is of the order of 0.1°. Analysis of flight images shows that potential sources of false detections, such as small dim clouds and sun glares, are correctly removed.

Challenges and Solutions for Vision-based Sense and Avoid

This paper focuses on sensing algorithms for vision-based non cooperative sense and avoid. Obstacle detection and tracking is based first on morphological filtering and local image analysis (detection), then on multi-frame processing in stabilized coordinates (tentative tracking), and finally on template matching and Kalman filtering-based state estimation (firm tracking). A conflict detection logic is introduced which uses an adaptive line-of-sight rate threshold based on the functional dependencies of the distance at closest point of approach in near collision conditions. The derived threshold takes into account ownship motion and estimated intruder azimuth, while assumptions are made regarding detection performance of the electro-optical system and intruder speed. The developed techniques have been tested using flight data gathered in a sense and avoid research project carried out by the Italian Aerospace Research Center and the Department of Industrial Engineering of the University of Naples “Federico II”. Achieved experimental results are promising and are discussed focusing on algorithm tuning and system performance in terms of probability of intruder declaration as a function of range, false alarm rate, tracking accuracy, and reliability of vision-based conflict detection.

Multi agent path planning strategies based on Kalman Filter for surveillance missions

The paper is based on the definition of a path planning strategy for surveillance missions with a system of multiple Unmanned Aircraft by means of a Kalman Filter technique. The developed method aims at finding a set of commands for the network of aircraft able to minimize a cost function whose definition depends on the mission. The approach adopted in this paper comprises several steps. The first one is based on the development of a target tracking algorithm to provide information on both target and drones motion on the surveillance area by means of a Kalman Filter and Bayesian network. Then, the objective functions can be defined depending on the relative position between aircraft and target. Finally, a heuristic approach allows finding the set of commands for the aircraft deployment over the surveillance area that maximize the utility function during the mission. The results demonstrate the ability of the tracking algorithm to provide accurate estimate of the target motion and the good capability of the whole system to react to the Command centre inputs based on the defined utility functions and decision making strategy.

Particle Filtering for Obstacle Tracking in UAS Sense and Avoid Applications

Obstacle detection and tracking is a key function for UAS sense and avoid applications. In fact, obstacles in the flight path must be detected and tracked in an accurate and timely manner in order to execute a collision avoidance maneuver in case of collision threat. The most important parameter for the assessment of a collision risk is the Distance at Closest Point of Approach, that is, the predicted minimum distance between own aircraft and intruder for assigned current position and speed. Since assessed methodologies can cause some loss of accuracy due to nonlinearities, advanced filtering methodologies, such as particle filters, can provide more accurate estimates of the target state in case of nonlinear problems, thus improving system performance in terms of collision risk estimation. The paper focuses on algorithm development and performance evaluation for an obstacle tracking system based on a particle filter. The particle filter algorithm was tested in off-line simulations based on data gathered during flight tests. In particular, radar-based tracking was considered in order to evaluate the impact of particle filtering in a single sensor framework. The analysis shows some accuracy improvements in the estimation of Distance at Closest Point of Approach, thus reducing the delay in collision detection.