Meet EBACE2019 Panelist Josh Gelinske, Appareo
13 May 2019
Josh Gelinske, director of AI systems for Appareo, will take part in a panel discussion on Wednesday 22 May about artificial intelligence and business aviation.
Titled, “Lexi, book me a jet,” the session will examine how the technology can be used in business aviation and include insights from industry representatives on the improvements AI contributes to the aviation sector.
In addition to Gelinske, panelists include Peter Conrardy, of GE Aviation; Bernhard Fragner, of GlobeAir and Kurt Doughty, from Collins Aerospace. The session will be moderated by Eric Leopold, director, Transformation for Financial and Distribution Services, International Air Transport Association.
Below is a Q&A with Gelinske to learn more about him and what he will discuss during his panel, which takes place from 15:00-16:00 in the Innovation Zone.
Q. Mr. Gelinske, could you briefly introduce yourself?
I am the director of AI systems for Appareo, where I lead a team of researchers and engineers focused on artificial intelligence and advanced software algorithms. We work with machine learning and deep learning to process information, as well as human-machine interaction through tools like augmented reality and 3D visualization. We then pull those aspects together to determine how such data may be presented to the consumer – in this case, the flight crew – to advise or assist them, and potentially suggest a course of action.
Q. You’ve been asked to participate in an EBACE session focused on artificial intelligence applications in the business aviation industry. What is your organisation’s involvement in this area?
Appareo is very active in potential AI applications within the aviation space. One key focus area our team has been working on is real-time speech recognition of air traffic control [ATC] transmissions. The system listens to communications between ATC and pilots, along with other transmissions such as weather broadcasts and airport information and turns that voice data into textual information presented to the flight crew on a mobile device like an iPad.
This helps increase the crew’s situational awareness by providing additional assistance to the pilot when they are focused on flying the aircraft. These sorts of capabilities are particularly useful to single pilot operators, and rotorcraft pilots who frequently have their hands occupied with the aircraft controls.
We’re also looking at even more interesting features to help pilots with this technology, such as incorporating national language processing capabilities to determine which audio transmissions are most relevant to the flight crew when operating in congested airspace with a lot of radio traffic.
Q. What do you hope to accomplish by participating in this panel?
I’m looking forward to a good discussion about the industry’s perceptions of AI. Aviation is obviously a high-tech industry, but it’s also in some ways a little slower to adopt new technologies due to concerns over safety and system robustness. I also expect some degree of naysaying, concern and risk aversion about what we’re looking to do, and I think that’s totally fair.
AI has been around a long time, but it’s only in the past five-seven years that it’s come out of academia and the labs to become useful to industry. We obviously must be careful in applying this technology, and we’ve seen some mistakes made when applying AI to other industries.
That presents an interesting dilemma as we look towards applying AI-based technologies to very complicated tasks in which humans often make mistakes. We’ve set the bar to where AI cannot be permitted to make mistakes, and we need to come to grips with that. Obviously, we’re looking to come as close to perfect as possible, but it’s unrealistic to expect a machine to be 100 percent infallible when a human is not.
Q. How do you think the business aviation community has evolved in the last 12 months? Any particular challenges you want to highlight?
Building on my earlier comments about user expectations, we may draw a parallel between future aviation applications and examples we’ve seen in other industries, such as pushing AI systems to ground vehicles. That has led to some mistakes that provide lessons on what should be done differently in aviation. For example, marketing AI feature sets as assistants helps to establish boundaries and ensures users don’t develop an impression the vehicle is in control more than it actually is.
We’ve also seen recent examples in which capabilities of traditional software systems have also become very high, which has led to incidents that have added fuel to people’s safety concerns. If we have such struggles with relatively traditional systems like angle of attack sensors, what issues will we encounter with next-level technologies like AI?
We must take the time to make sure we do not employ such systems until they are truly capable of assuming those tasks. Autonomous systems are data-driven, and there are a lot of steps involved between identifying an application for AI and implementation. If we’re to enable certain AI features five years from now, we need to get the ball rolling today and start gathering that data to start building towards that goal.
Q. Take our three-word challenge: pick three words to describe the future business aviation industry.
Intelligence, safety and performance.