Applying Sensor Uncertainty Mitigation Schemes to Detect-and-Avoid Systems

Michael Abramson; Mohamad Refai; M Gilbert Wu; Seungman Lee

NASA NTRS · 20205002286

Applying Sensor Uncertainty Mitigation Schemes to Detect-and-Avoid Systems

Michael Abramson·Mohamad Refai·M Gilbert Wu·Seungman Lee

Abstract

Impact of sensor noise on the performance of Detect-And-Avoid (DAA) systems can be reduced by implementing various mitigation schemes. This paper evaluates the Sensor Uncertainty Mitigation (SUM) method, implemented in the Detect and Avoid Alerting Logic for Unmanned Systems (DAIDALUS) algorithm, a reference implementation in the DAA minimum operational performance standards. DAIDALUS SUM performance is evaluated using a few safety and operational suitability metrics and compared with more traditional approaches using static safety buffers. A large number of encounters representative of low-speed unmanned aircraft against non-cooperative manned aircraft are simulated and evaluated. An air-to-air radar model produces representative sensor noise for the DAA system. Results show that increasing the tunable parameters for horizontal and vertical uncertainty in DAIDALUS SUM improves the safety metric at the cost of increasing the number of system alerts leading to increased workload. A range of SUM parameters is recommended as suitable values for the type of operations considered for this work. General trends and optimal SUM configurations were found to be nearly the same for two large and very different encounter data sets.

Explore related subjects

detect and avoid uncertainty mitigation simulation

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Michael Abramson, Mohamad Refai, M Gilbert Wu, Seungman Lee. Applying Sensor Uncertainty Mitigation Schemes to Detect-and-Avoid Systems. https://ntrs.nasa.gov/citations/20205002286

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

NASA NTRS2018

Utility of Visual and Auditory Warning Alerting for Traffic Avoidance During UAS Operations

This presentation covers the results of a human-in-the-loop simulation that examined the effect of caution versus warning-level alerting in the context of Unmanned Aircraft System (UAS) operations in civil airspace. Minimum Operational Performance Standards (MOPS) are currently being developed by RTCA Special Committee 228 for UAS Detect-and-Avoid (DAA) systems. Input from subject matter experts and multiple research studies have informed display requirements for DAA systems aimed at supporting timely and appropriate pilot responses to collision hazards. DAA alerting is designed to inform pilots of potential threats to DAA "well clear"; the two highest alert levels - caution and warning - indicate how soon pilot action is required and whether there is adequate time to coordinate with the air traffic controller (ATC). Additional empirical support is needed to clarify the extent to which warning-level alerting impacts DAA task performance. The present study explores the differential effects of the auditory and visual cues provided by the DAA Warning alert, and performance implications compared to caution-only alerting are discussed.

detect and avoid↗

NASA NTRS2017

UAS Well Clear Recovery Against Non-Cooperative Intruders Using Vertical Maneuvers

This paper documents a study that drove the development of a mathematical expression in the minimum operational performance standards (MOPS) of detect-and-avoid (DAA) systems for unmanned aircraft systems (UAS). This equation describes the conditions under which vertical maneuver guidance could be provided during recovery of well clear separation with a non-cooperative VFR aircraft in addition to horizontal maneuver guidance. Although suppressing vertical maneuver guidance in these situations increased the minimum horizontal separation from 500 to 800 feet, the maximum severity of loss of well clear increased in about 35 of the encounters compared to when a vertical maneuver was preferred and allowed. Additionally, analysis of individual cases led to the identification of a class of encounter where vertical rate error had a large effect on horizontal maneuvers due to the difficulty of making the correct left-right turn decision: crossing conflict with intruder changing altitude. These results supported allowing vertical maneuvers when UAS vertical performance exceeds the relative vertical position and velocity accuracy of the DAA tracker given the current velocity of the UAS and the relative vertical position and velocity estimated by the DAA tracker. Looking ahead, these results indicate a need to improve guidance algorithms by utilizing maneuver stability and near mid-air collision risk when determining maneuver guidance to regain well clear separation.

detect and avoid↗

NASA NTRS2017

The Generic Resolution Advisor and Conflict Evaluator (GRACE) for Detect-And-Avoid Systems

Java Architecture for Detect-And-Avoid (DAA) Extensibility and Modeling (JADEM) was developed at NASA Ames Research Center as a research and modeling tool for Unmanned Aircraft Systems (UAS) Integration in the National Airspace System (NAS). UAS will be required to have DAA systems in order to fulfill the regulatory requirement to remain well clear'' of other traffic. JADEM supports research on technological requirements and Minimum Operational Performance Standards (MOPS) for UAS DAA systems by providing a flexible and extensible software platform that includes models and algorithms for all major DAA functions. This paper describes one of these algorithms, the Generic Resolution Advisor and Conflict Evaluator (GRACE). GRACE supports two core DAA functions: threat evaluation and guidance. GRACE is generic in the sense that it is designed to work with any aircraft or sensor type (both cooperative and non-cooperative), and to be used in various applications and DAA guidance concepts, thus supporting evolving MOPS requirements and research. GRACE combines flexibility, robustness, and computational efficiency. It has modest memory requirements and can handle multiple cooperative and noncooperative intruders. GRACE has been used as a core JADEM component in several real-time and fast-time experiments, including human-in-the-loop simulations and live flight tests.

detect and avoid↗