Is the effect of aviation automation on pilots comparable to the effect vehicle automation on drivers? Decades of research has focused on the human factors implications of automating different aspects of aviation. However, much of the human factors research into the field of vehicle automation seems to not be drawing on this experience either in their literature or in the design of the methodologies and I wondered why. Is it that researchers are unfamiliar with aviation automation literature or is there something fundamentally different between the two domains that make comparisons difficult?
Some less philosophical thoughts about this question:
Differences:
- Professional operators with tons of up-to-date licenses in aviation, laymen in road traffic with standard licenses from some decades ago.
- Research in aviation on this topic has started in the late 70s which was mostly public funded and nearly everything is available online and in English. Much automotive research has started these days, is funded or done by car manufacturers, everything needs publication approvals from the company’s project leader, his boss, his boss-boss and his boss-boss-boss, several vice-presidents and their grandmothers and much work is also done in German (often in parallel to the English-speaking community).
- Completely different vehicles, technical systems, environments, jurisdictions.
- In aircraft accidents the manufacturers are traditionally blamed for some design issues, in car accidents the manufacturers normally don’t get blamed.
- Many of us are really frightened of airplane crashes, which are perceived as other-directed risks, which leads to an over-estimation of these risks, while we usually evaluate the risks of car accidents as self-determined and so we underestimate these risks.
- Many aviation researchers do not fly by themselves, while nearly everybody owns or drives a car.
Similarities:
- Human operators with the same sensory and cognitive systems in both cases! In both cases they belief to perform better than they do.
- I predict the same automation effects we have already seen in aviation to happen in the automotive domain like reduced vigilance and loss of situation awareness on short term and skill degradation on long term.
- Needed disciplines, competences, equipment, and methods for research are mostly the same.
It is my impression that Human Factors scientists are generally quite familiar with the aviation literature. The first autopilots and flight simulators were introduced before WW2, and it can be said that many of the theories/knowledge we use are based on previous experiences in aviation and air traffic control (Sheridan, 2002). A key difference between flying and driving is that flying is highly proceduralized with long time constants, while in driving there many interactions with (vulnerable) road users and tight time margins.
Fitts, P. M. (Ed.) (1951). Human engineering for an effective air-navigation and traffic-control system. Washington, DC: National Research Council.
Sheridan, T. B. (2002). Humans and automation: System design and research issues. John Wiley & Sons, Inc..
Wheeler, W. A. & Trigs, T. J. (1996). A task analytical view of simulator based training for drivers. Proceedings of the Road Safety Research and Enforcement Conference ‘Effective Partnerships’, Coogee Beach, New South Wales (NSW), Australia, 4-5 November 1996, 217–221.
Hi Tyron,
You may refer to Young, Stanton & Harris (2007) who specifically addressed this question (attached). I am pretty sure that Neville Stanton has published other papers on the same topic.
You raise an interesting question and I think the answer is also partially found in the concept of "swim lanes". In my judgment, it seems that after people have worked and published within a certain domain, or "swim lane" (e.g. Human Factors vs. Aviation), they fall into a routine of only regarding research within that domain as well as delimiting their publication venues to the journals and outlets that fall within their "swim lane". The way in which I attempt to prevent myself from doing this same thing is by using as broad search terms as possible when seeking research on a particular topic within my domain (e.g. Trust in Automation, Human-System Integration, Situational Awareness, etc). It regularly happens that I find and read relevant works that are not typically cited in the published works in my domain. This is because researchers in other related domains (e.g. Systems Engineering, Military Psychology, Computer Science, etc) tend to investigate similar problems but publish using a slightly different lexicon - including identified keywords and so forth. I also find it helpful to dissect the phenomena that I study into very small, "elemental" components (such as value-based decision making, motor planning, object detection, etc) and search in the basic science literature as well as in the more applied human factors and ergonomics domain.
That said, I also agree with those above who suggest that those in the main Human Factors literature do indeed seem well aware of what has been done in the context of aviation Automation. For instance in the work of the late Raja Parasuraman (e.g. Parasuraman & Riley, Human Factors, 1997), it was common to find reference to automation of broad types (aviation, railroad, military, etc).
Hi Jason,
Thanks for your input! I know that aviation is referred to in position/concept/high-level papers on automation, but even if it is the case that most HF researchers are aware of the aviation literature I find that it is used sparingly in the preamble to empirical work on similar issues in vehicle automation, except for perhaps mentioning that similar problems have been faced in aviation (cue mandatory Parasuraman reference).
Of course, the two tasks are structurally unique (as are their contexts) and, as Joost suggests, driving is arguably a much more complex task. However, I wonder, and I've not delved into the deep end here myself, whether when we are thinking about all manner of HMI/design solutions to simple interaction problems, whether the decades of HF research on human-automation-interaction in aviation might not give us some ideas how to tackle parallel problems in vehicle automation. Perhaps they already are and people just aren't saying!
Tyron,
I agree with your perspective. Indeed, I think that most domains would do well to look deeply and better integrate the valuable information, findings, and solutions from those who have approached the same or highly similar problems spaces previosly.
At the same time, I think that these works may have been influential, even if not overtly or explicitly referenced at times. Because, economy of written presentation and so forth . . .
As far as the contexts being different, I do not have much to add beyond that which Joost said above. Control in flight seems more complex simply by adding the need to monitor/control altitude when that is not available in terrestrial transportation. At the same time, automation has made its way into aviation to a much greater degree (and in a variety of ways simplified control), at a minimum, because there seem fewer immediate threats that must be considered (commercial planes are not going to "run off the road" or hit a tree).
Either way, I think you pose both an interesting question and challenge to those of us who work in this domain. Funny that, I've been working with a colleague to co-author a current literature review focused on the question of whether trust in automation can be calibrated (or leveraged) in real-time to facilitate improved human-automation interaction. We were literally just discussing how broad the review should reach and, specifically, I was challenging her attempt to couch many arguments in reference to the UAV literature when we, ourselves, work in the domain of driving automation. Thus, your challenge is timely and relevant to work that I am likely to submit in the near future!
Thank you for initiating this discussion. Hopefully others read it and are likewise encouraged to be broad in their awareness of how their research relates to the world of science going on around them.
Cheers!
Some less philosophical thoughts about this question:
Differences:
- Professional operators with tons of up-to-date licenses in aviation, laymen in road traffic with standard licenses from some decades ago.
- Research in aviation on this topic has started in the late 70s which was mostly public funded and nearly everything is available online and in English. Much automotive research has started these days, is funded or done by car manufacturers, everything needs publication approvals from the company’s project leader, his boss, his boss-boss and his boss-boss-boss, several vice-presidents and their grandmothers and much work is also done in German (often in parallel to the English-speaking community).
- Completely different vehicles, technical systems, environments, jurisdictions.
- In aircraft accidents the manufacturers are traditionally blamed for some design issues, in car accidents the manufacturers normally don’t get blamed.
- Many of us are really frightened of airplane crashes, which are perceived as other-directed risks, which leads to an over-estimation of these risks, while we usually evaluate the risks of car accidents as self-determined and so we underestimate these risks.
- Many aviation researchers do not fly by themselves, while nearly everybody owns or drives a car.
Similarities:
- Human operators with the same sensory and cognitive systems in both cases! In both cases they belief to perform better than they do.
- I predict the same automation effects we have already seen in aviation to happen in the automotive domain like reduced vigilance and loss of situation awareness on short term and skill degradation on long term.
- Needed disciplines, competences, equipment, and methods for research are mostly the same.
As an "old timer" in HF, I like to learn by Jason's comments that also young researches are aiming rersearch goals not with a tunnel perspective but with a broader consideration of different aspects - just that what our brain is doing when faced with a new situation: fusion of all information available in order to determine an optimized response.
Why the research community did somehow loss the broad-band-fusion of information in most publications?
- People try to publish as many papers as possible for career aspects; this leads to short papers which provide no space for discussing far-related aspects.
- Reviewers do not invest much time in a thorough judgement of manuscripts. Even experienced, I always need around 3 full days for a review to check the content and all surounding relationsships of a manuscript, like Tyron expect it in his initial question. Therefore, I am serving 1 review request per month only, which is 20 percent of the incoming requests. My observation is that editors escape to young PhD students (having published a few papers only, and have no time to spend much energy in preparing a review) to obtain the necessary amount of reviews. This bad situation can be antagonised by showing with every publication the names of the reviewers; then, reviewers take public responsibility tor the content of the publication.
Tyron, thank you very much for initionalising this discussion! Jason, hopefully your research mind will find many followers!
A clue: If you analyze traffic in an airport and traffic in a road, you will see that road should be unmanageable using the airport approach.
Air space is mainly empty space and, even without a control, chances for midair collisions are low. In other words, the main task for automation in Aviation is handling the plane. Driving a car is different: The main task will be the interaction with other cars, people and so on. This single fact makes automation much harder in cars than in planes.
Tyron,
I am an Air Force fellow at the US Dept of Transportation, and your question is my research topic exactly. I am an Air Force pilot (KC-10s) by background and have the opportunity to work with the Department on strategic policy regarding Automation in vehicles. I will be doing a comparative study between automation in aviation and that in cars, mainly an analytical piece focused on policy of V2V/V2I technologies, through the lens of aviation automation policy. I was just curious if you have found anything significant since your original post, regarding how these industries have collaborated, and if not, why not?
Best,
Christina
Great question. One difference is that automation research in aviation was largely driven by government (esp. military) problems, whereas automation in ground vehicles is being driven largely by commercial interests. I have noticed many vehicle manufacturers creating or adopting technologies that don't seem that grounded in the human factors science, perhaps because they need to get the next big thing to market and don't have the luxury of waiting years for the studies to be finished. I could be wrong, b/c the $$ spent on R&D is increasing for industry in proportion to govt. lately.
- Badges
- Science topic
- Similar topics
- Psychology