Recently, Embry-Riddle and Florida Polytechnic University entered into a research partnership (press release) where safety research initially developed for autonomous vehicles (AVs) will be applied to aerospace applications. This is a bit surprising given the rich history of autonomous systems in aerospace. (Full Disclosure: I initiated and built the Advanced Mobility Institute at Florida Polytechnic University)
As early as 1914, Lawrence Sperry demonstrated the concepts of automated flight through an interlock process which would attempt to keep the “programmed heading.” Since this time, airborne systems have increased in sophistication with the use of electronic hardware and software. Important system innovations in the aerospace sector have included clearly defining the operating decision domains (ODDs) such as taxi, takeoff, climb, cruise, descent and landing. Aerospace has also focused on redundancy for the sensor systems, and a regulatory framework to use the capabilities in important functions such as instrument aided landings (ICAO).
Today, the use of automated technologies is so prevalent in aerospace that it is no longer news. Overall, the aerospace ecosystem has been a great success, and with this success, there has been a call to apply aerospace technologies to the automotive sector. Of course, there is much to learn. AVs are starting to use ODD concepts such as geo-fencing. Also, if the aerospace sector is an indication, a gradual and safe deployment creates a conducive environment for consumer acceptance. However, it is important to note the significant differences between aerospace and automotive.
Table 1: Comparison of autonomous challenges for aerospace and automotive
As Table 1 indicates, while the inherent safety stability point for aerospace is more challenging, automotive must operate in a much denser environment. There are more cars in motion in a mid-sized city than in the entire airspace of the united states. Further, these cars are operating in a space which has one less dimension and only on paved surfaces. This leads to orders-of-magnitude higher density and the associated lack of margin.
What is an example of this density ? Consider radar as a tool. Radar is a very useful in aerospace because all the objects one detects via radar are of interest. However, in automotive, the environment is so congested and the reflections so complex, that radar as a tool is much less effective. Other issues include operating in an actively managed traffic environment with limited degrees of freedom, close-order engagement with civil infrastructure (bridges, tunnels, work zones) and of course, the most challenging being the close-order engagement with humans. Thus, automotive safety science is a very challenging problem.
Interestingly, the future of aviation is contemplating use-models such as drones for last mile delivery — in an environment which most closely resembles the traditional automotive environment. Even more exciting are short-hop taxi services such as lilium which offer the promise of efficient transport above the constraints of the existing transportation system. Each of these solutions delves into many of the same issues faced by the current AV technologies and the potential for a very interesting cross-pollination of fields.
Note: A deeper examination of the details of AV Safety exists in a Society of Automotive Engineers (SAE) report which can be found at SAE EDGE Research Reports.
