Conflict Probe Testing: Using Biology Inspired Genetic Algorithm
By Linda Weiland, Embry Riddle University
According to James Ritchie III, recent Rowan University graduate, FAA intern, and soon to be full-time Tech Center employee, biology and aviation have a lot in common. He believes scientists can draw on the study of biology to assist in automation in air traffic, and that selection (think “survival of the fittest”) of time-shifted modules can be built with scenarios to avoid conflict, in both horizontal and vertical separation, in all phases of flight and encounter angle.
Ritchie, along with three members of the FAA, published a paper titled, “Design and Performance of an Improved Genetic Algorithm Implementation for Time- Shifted Air Traffic Scenario Generation.” He presented his paper on Aviation STEM Monday and Tech Center Tuesday and focused on conflict probe testing that mimics biological evolution.
NextGen’s trajectory-based operations (TBO) requires enhanced automation, conflict resolution, and Decision Support Tools (DST) to assist controllers in managing air traffic. Ritchie’s research focused on a DST called Conflict Probe (CP), which predicts future conflicts in air traffic. CP depends on performance, and performance must be quantified. The key metrics used were missed alerts, valid alerts, false alerts, and correct no-calls from operational performances.
To obtain a quantitative evaluation the work centered around an empirical testing (numbers) process with recorded or simulated traffic data. With recorded traffic data, the team was able to develop time-shifted traffic scenarios and used biological evolution and its theory of survival of the fittest to obtain the values for the time-shifts. Originally random generated time-shift values for each flight were used, but this produced a lack of control on conflict characteristics in flight. The questions became, “Can we do better? And how do we find a faster yet safe method of time-shifting?” This is where the team turned to biology. The team reviewed past studies such as research on the movement of birds’ wings and the movement of whales’ fins to find solutions for technology challenges. These genetic algorithms, which use concepts from evolution, helped the team synthesize “best solutions" for problems that have many constraints. He likened airplanes to genes and planes traveling in a single airspace to chromosomes. This metaphor helped build the Genetic Algorithm Structure that could be evaluated, would show fitness, and identify conflict characteristics. The more criteria that could be met (horizontal and vertical separation, encounter angle, and phase of flight), the higher the fitness.
Once selected as “fit,” one chromosome (a group of planes) could be chosen to mate with another "fit" chromosome (a separate group of planes), and once technology had mated these so-called chromosomes, the parent groups could produce offspring with swapped genetic material. Once there were two-point crossovers, researchers would have time-shift values (planes in the right place).
Ritchie summed it up best at the end of his extended abstract, “the improved implementation combined with the knowledge gained from these experiments will provide the FAA a valuable tool in generating future air traffic scenarios for testing air traffic control automation."
Talking to Ritchie after the Tech Talk, he shared that he did his first internship in high school, which started the ball rolling to his upcoming full-time employment.
“Internships that also include good mentors is very important to the evolution of technology,” says Ritchie. He takes this thought literally, and he is speaking to others students about getting involved in an internship program.