CASE STUDY
iDAS
Instructors sometimes find it hard to quantify what is causing student behaviour – they can see the effect – for example the aircraft speed dropping – but it is hard to assess if the student is aware of what is causing the problem – ie have the students checked horizon/their instruments for attitude and has the instructor actually seen what the student was concentrating on when the error was made.
Cause/effect need to be correlated and instructors don’t always get it right. Being able to quantify the actions they take in order to assess and correct a problem would be of benefit in an after action review.
-Senior RAF Instructor
The challenge we set ourselves on Project iDAS was to understand this idea of cause and effect in fast jet pilots, particularly in relation to Airmanship in RAF pilots.
Airmanship is typically a subjective assessment of the students performance, made by an instructor. We are aiming to use data to give instructors a tool to allow them to provide a more objective assessment.
Challenge
This data is currently being analysed and initial results are providing some very interesting insights. As this data set is too large for a human to comprehend, we are currently developing and deploying machine learning across it in order to better understand how we can codify airmanship behaviours. We are also currently expanding the iDAS data capture system to work with non-VR applications
Result
.png)
We combined commercial off the shelf flight simulation technology, both software and hardware, with a VR headset and HEAT.
The Project iDAS team included leading academic experts in behavioural psychology and machine learning. Working with these experts we devised a number of scenarios to test lookout and situational awareness characteristics in pilots. We also identified what we believed would be the key data points for artificial intelligence to understand what the pilots were doing.
We build all of this knowledge into a man portable system which can be carried on commercial airlines and set up at a standard office desk in approximately 15 minutes.
We used this portability to travel to 3 active duty RAF stations and capture data from 39 active duty pilots. In total we captured almost 1 billion data points
Solution
Instructors sometimes find it hard to quantify what is causing student behaviour – they can see the effect – for example the aircraft speed dropping – but it is hard to assess if the student is aware of what is causing the problem – ie have the students checked horizon/their instruments for attitude and has the instructor actually seen what the student was concentrating on when the error was made.
Cause/effect need to be correlated and instructors don’t always get it right. Being able to quantify the actions they take in order to assess and correct a problem would be of benefit in an after action review.
-Senior RAF Instructor
The challenge we set ourselves on Project iDAS was to understand this idea of cause and effect in fast jet pilots, particularly in relation to Airmanship in RAF pilots.
Airmanship is typically a subjective assessment of the students performance, made by an instructor. We are aiming to use data to give instructors a tool to allow them to provide a more objective assessment.
Challenge
We combined commercial off the shelf flight simulation technology, both software and hardware, with a VR headset and HEAT.
The Project iDAS team included leading academic experts in behavioural psychology and machine learning. Working with these experts we devised a number of scenarios to test lookout and situational awareness characteristics in pilots. We also identified what we believed would be the key data points for artificial intelligence to understand what the pilots were doing.
We build all of this knowledge into a man portable system which can be carried on commercial airlines and set up at a standard office desk in approximately 15 minutes.
We used this portability to travel to 3 active duty RAF stations and capture data from 39 active duty pilots. In total we captured almost 1 billion data points