James K. Kuchar

A review of conflict detection and resolution modeling methods

A number of methods have been proposed to automate air traffic conflict detection and resolution (CDR), but there has been little cohesive discussion or comparative evaluation of approaches. The paper presents a survey of 68 CDR modeling methods, several of which are currently in use or under operational evaluation. A framework that articulates the basic functions of CDR is used to categorize the models. The taxonomy includes: dimensions of state information (vertical, horizontal, or three-dimensional, 3-D); method of dynamic state propagation (nominal, worst case, or probabilistic); conflict detection threshold; conflict resolution method (prescribed, optimized, force field, or manual); maneuvering dimensions (speed change, lateral, vertical, or combined manoeuvres); and management of multiple aircraft conflicts (pairwise or global). An overview of important considerations for these and other CDR functions is provided, and the current system design process is critiqued.

Prototype Conflict Alerting System for Free Flight

The development of a prototype alerting system for a conceptual free eight environment is discussed. The alerting logic is based on a probabilistic model of aircraft sensor and trajectory uncertainties that need not be Gaussian distributions. Monte Carlo simulations are used over a range of encounter situations to estimate coneict probability as a function of intruder position, heading, and speed, as determined through a datalink between aircraft. Additionally, the probability of coneict along potential avoidance trajectories is used to indicate whether adequate space is available to resolve a coneict. Intruder intent information, e.g., e ight plan, is not included in the modelbutcould beused toreducetheuncertaintyintheprojected trajectory. Fouralertstagesaredeenedbasedon the probability of cone ict and on the avoidance maneuvers that are availableto thee ight crew. Preliminary results from numerical evaluations and from a piloted simulator study at NASA Ames Research Center are summarized.

Methodology for Alerting-System Performance Evaluation

A probabilistic-analysis methodology is described that provides quantitative measures of alerting-system performance, including the probabilities of a false alarm and missed detection. As part of the approach, the alerting decision is recast as a signal-detection problem, and system operating-characteristic curves are introduced to describe the tradeoffs between alerting-threshold placement and system performance. The methodology fills the need for a means to determine appropriate alerting thresholds and to quantify the potential benefits that are possible through changes in the design of the system. Because the methodology is developed in a generalized manner, it can be used in a variety of vehicle, transportation-system, and process-control applications. The methodology is demonstrated through an example application to the Traffic Alert and Collision Avoidance System (TCAS). Recent changes in TCAS alerting thresholds are shown to reduce the probability of a false alarm in situations known to produce frequent nuisance alerts in actual operations. Nomenclature A, N, T = probabilistic-state trajectories E = event of encountering a hazard fx (x) = probability density function for the random variable x h = estimated relative altitude h = estimated relative-altitude rate / = incident event ^maif - probability of alerting-system malfunction P(x) = probability of event ;c PT(X) = probability of event x evaluated along trajectory T r = estimated relative range r = estimated relative-range rate

Airspace Encounter Models for Estimating Collision Risk

Airspace encounter models, providing a statistical representation of geometries and aircraft behavior during a close encounter, are required to estimate the safety and robustness of collision avoidance systems. Prior encounter models, developed to certify the Traffic Alert and Collision Avoidance System, have been limited in their ability to capture important characteristics of encounters as revealed by recorded surveillance data, do not capture the current mix of aircraft types or noncooperative aircraft, and do not represent more recent airspace procedures. This paper describes a methodology for encounter model construction based on a Bayesian statistical framework connected to an extensive set of national radar data. In addition, this paper provides examples of using several such high-fidelity models to evaluate the safety of collision avoidance systems for manned and unmanned aircraft.

Survey of conflict detection and resolution modeling methods

The design and evaluation of traffic conflict detection and resolution systems requires the use of analytical models that describe encounter dynamics and the costs and benefits of avoidance actions. A number of such models have been applied in the past to the problem, but there has been no cohesive discussion or comparative evaluation of these approaches. Each method has benefits and limitations, and future efforts may be facilitated by combining the best features of different techniques. This paper presents a summary of conflict detection and resolution modeling approaches. Modeling techniques are categorized and the fundamental assumptions, capabilities, and limitations of each approach are described. The methods are evaluated and compared based on their applicability to free flight traffic conflict situations.

Probability-based collision alerting logic for closely-spaced parallel approach

A prototype airborne collision alerting logic was developed for aircraft on approach to closely-spaced parallel runways. A novel design methodology was used based on collision probabilities instead of traditional spatial or temporal alerting criteria. With this technique, an alert is issued when the probability of a collision exceeds an acceptable threshold value. The logic was based on a hazard level corresponding to the current Precision Runway Monitor System of one accident in every one thousand worst-case blunders. Probability contours were constructed through Monte Carlo simulation over a range of aircraft position, speed, heading, and turn rate conditions. These contours were stored in look-up tables that were accessed in real time for evaluation during numerical simulation of approaches. Three runway spacings were investigated: 3,400, 2,500, and 1,700 ft. The results show that the unnecessary alert rate at 1,700 ft runway spacing was double that at 3,400 ft runway spacing. Additionally, the logic induced collisions in two low-closure-rate situations, suggesting that the worst case blunder may not be a drastic heading change. (Author)

Collision Avoidance for Unmanned Aircraft using Markov Decision Processes

we investigate the automatic generation of collision avoidance algorithms given models of aircraft dynamics, sensor performance, and intruder behavior. By formulating the problem of collision avoidance as a Markov Decision Process (MDP) for sensors that provide precise localization of the intruder aircraft, or a Partially Observable Markov Decision Process (POMDP) for sensors that have positional uncertainty or limited eld-of-view constraints, generic MDP/POMDP solvers can be used to generate avoidance strategies that optimize a cost function that balances ight-plan deviation with collision. Experimental results demonstrate the suitability of such an approach using four dierent sensor modalities and a parametric aircraft performance model.

A Real-Time Monte Carlo Implementation for Computing Probability of Conflict

*† ‡ § In this paper, we present a method for computing the Probability of Conflict, PC, using a fast Monte Carlo implementation. Often, Monte Carlo simulations are associated with offline analysis or verification, and are thought of as too slow for real-time usage. However, we describe an implementation that allows for fast computation and can be used in certain online applications. Examples are given for aircraft and automobile conflict detection. Both pseudo-code and Matlab code are provided to help explain and disseminate the approach.

Evaluation of Collision Avoidance Maneuvers for Parallel Approach

Two candidateevasion maneuverswere evaluated foruse with acollision alerting systemforindependent closely spaced parallel approaches in instrument conditions. The two maneuvers were a wings-level climb and a climbing turn away from parallel trafe c. Pairs of aircraft on parallel approach were simulated by use of prerecorded trajectories covering a range of normal approach and blunder examples. Each example was repeated twice, with the endangered aircraft responding to alerts with either the climb-only or the climbing-turn evasion. The climbonly maneuver is shown to result in 38-times as many collisions as the climbing-turn for nominal alert threshold settings. It is possible to reduce the collision rate by adjusting threshold parameters, but the false alarm rate increases. The climb-only maneuver is shown to be uniformly less safe than the climbing turn for all parameter combinations. Results are illustrated with system operating characteristic curves.

Preliminary study of behavioral and safety effects of driver dependence on a warning system in a driving simulator

Warning systems are being developed to improve traffic safety using visual, auditory, and/or tactile displays by informing drivers of the existence of a threat in the roadway. Behavioral and safety effects of driver dependence on such a warning system, especially when the warning system is unreliable, were investigated in a driving-simulator study. Warning-system accuracy was defined in terms of miss rate (MR) and positive predictive value (PPV) (PPV is the fraction of warnings that were correct detections). First, driver behavior and performance were measured across four warning-system accuracy conditions. Second, the authors estimated the probability of collision in each accuracy condition to measure the overall system effectiveness in terms of safety benefit. Combining these results, a method was proposed to evaluate the degree of driver dependence on a warning system and its effect on safety. One major result of the experiment was that the mean driving speed decreased as the missed detection rate increased, demonstrating a decrease in drivers reliance on warnings when the system was less effective in detecting threats. Second, both the acceleration-pedal and brake-pedal reaction times increased as the PPV of the warning system decreased, demonstrating a decrease in driver compliance with warnings when the system became more prone to false alarms. A key implication of the work is that performance is not necessarily directly correlated to warning-system quality or trends in subjective ratings, highlighting the importance of objective evaluation. Practical applications of the work include design and analysis of in-vehicle warning systems