(We originally posted this in 2020. You can read more of our original ideas in our archive.)
Problem: Air Traffic Controlling has a variety of problems at scale that has been explored quite in depth through the research (as described by Lucio Bianco Maurizio Bielli in their paper “Air Traffic Management: Optimization Models and Algorithms”):
The presence of several human operators in the system (pilots, controllers, etc.), each having a certain independence in making decision;
A high number of variables and constraints; numerous sub-systems and strong interactions;
Several control objectives also conflicting;
Limitations and complexity of models used to forecast the movement of airplanes and traffic evolution;
Fast dynamics and real-time interventions
Solution: Above are some visual representations of the air traffic control system as modeled out from this great paper which I also referenced above. This research has also been built upon by Claire Tomlin (berkeley.edu): some of the highlights include her classes Hybrid System Theory (EE291E) (spring 2009-2018) and her Hybrid Systems Lab. As described on their site, “These are systems which combine continuous time dynamics with discrete event dynamics. Research involves design, verification using new computational methods for computing reachable sets, and simulation of hybrid systems. We are currently developing the theory and designing models to predict the behavior of complex systems. This work is supported by NSF, ONR, and AFOSR.”
This business would take the types of research that Tomlin works on to innovate on ways to create automated air traffic controllers. One area where this business would thrive, for instance, is in the world of turbulence. Currently pilots know where turbulence is because of knowledge from air traffic controllers and past pilots. If one plane experiences turbulence, they let the air traffic controllers know who then map this and report it out to other pilots who may potentially fly through or around it. This could quite easily be done with an algorithm, even if the only resource to track this is radio.
The air traffic controller market is quite large: on average, they get paid $122k per year (based on 2019 Median Pay) and there are 24,300 globally (see Number of Jobs, 2019 or Job Outlook, 2019-29), costing about $3 billion a year. I predict that within a decade about 90% of these jobs can and will be replaced with machines or algorithms. Given the long lead-time and high requirement of safety, I also envision that businesses in this realm will have extremely high defensibility due to barriers-to-entry of competitors.
Currently, this industry is still heavily in the “research-phase” (see this paper by NASA or this paper by IEEE) but have virtually no corporate competitors.
Monetization: Fees for licensing or software/service usage.
Contributed by: Michael Bervell (Billion Dollar Startup Ideas)