Autonomy: UAS and Manned Aircraft

 Hello Again Fellow Aces, 

This week and the previous week a bulk of my research was on the subject of automated systems and autonomous operations by UAS and manned aircraft. My major research project which will be available in a post here in a few weeks (stay tuned) is specifically focused on the increased levels of automation and the benefits and drawbacks that can be attributed to its presence in aviation. 

According to the Parasuraman, Sheridan, and Wickens as cited by Sheridan (2018) in her dissertation, The Effect of Task Load, Automation Reliability, and Environment Complexity on UAV Supervisory Control Performance, there are 10 levels of automation. The levels start with fully manual operations and scale with increased examples of automation for each successive level.  For example, level 1 automation offers no assistance to the human who must take all actions and make decisions.  Level 2 offers a pool of possible decisions, and level 3 narrows down the selection of decisions to the more optimal alternatives. By level 4 the computer has narrowed the selection to just one option, but still allows the human to execute. At the top half of the scale, the computer makes a decision and then either awaits a human veto (Level 6) or executes automatically and informs the human of actions taken (level 7). For a complete description of all 10 levels see the table below.  

I think there are currently different considerations regarding the implementation of automation in manned aircraft versus UAS operations. These considerations are mostly shaped by public perception of the operations. In manned aircraft equipped with much of the same automation as UAS, there is typically more thought given to remaining in the middle of the scale (levels 4-7) to allow the human to still be in the loop on what the aircraft is doing.  UAS tend to lean toward higher levels of automation because of the operator's physical separation from the aircraft, limited perspective, and delay in flight control commands (especially when operating beyond line of sight). The tendency is to believe that automating many operations at a a higher level for the UAS enables safer operations.

In general, I think the trend towards building a more robust and reliable network that supports high levels of automation is beneficial, but new emerging phenomenon in the realm of human factors like automation surprise (AS) need to be carefully considered as higher levels of automation become commonplace.  More automation can enable highly efficient airspace operations and statistically provide a safer environment if working properly, so I would argue that more automation in the industry would be welcome as long as due diligence has been given to mitigate design characteristics that may be undesirable.

Fly Safe! 

-DP

References

Endsley, M. R. (2017). From Here to Autonomy: Lessons Learned From Human–Automation Research. Human Factors, 59(1), 5-27. 10.1177/0018720816681350

Etherington, T. J., Kramer, L. J., Le Vie, L. R., Last, M. C., Kennedy, K. D., Bailey, R. E., & Houston, V. (2018). Impact of Advanced Synoptics and Simplified Checklists during Aircraft Systems Failures. Paper presented at the 1-9. 10.1109/DASC.2018.8569559 http://erau.summon.serialssolutions.com/2.0.0

Sherwood, S. M. (2018). The Effect of Task Load, Automation Reliability, and Environment Complexity on UAV Supervisory Control Performance. https://commons.erau.edu/edt/434