Opportunities in autonomy: 
Royal Academy of Engineering 
debate 

Innovators
Giving our customers an operational advantage

Our Technology Director Dave Short took part in this debate with Professor Paul Newman, the founder and CTO of Oxbotica. It was chaired by Dame Muffy Calder, the former Chief Scientific Advisor for Scotland.

 
Image of Dave Short I hope you get chance to watch the debate and hear about some of the opportunities we discussed.
 
One point we talked about is developing trust in Artificial Intelligence and autonomy, which will need to happen hand-in-hand with its wider use. 
 
When we think about autonomy, there’s a big difference between deterministic and non-deterministic systems. Stay with me here, because it’s an important point that also involves how our own brains work.
 
In deterministic systems, a certain input will always lead to a certain action. The airbag on a car is deterministic autonomy, because we know it will be triggered when sensors detect a crash.
 
In non-deterministic systems, we can’t always predict a response based on the inputs, as there’s no strict if x then y response. To make an autonomous car for example, you need it to make decisions even though you can’t pre-define every situation it will face. Because of that, you don’t always know how it will react, which is why it’s so hard to prove it’ll be safe in all situations.
 
Unmanned P24 Mk IV  
Unmanned Pacific 24 Mk IV autonomous sea boat
 
There’s another non-deterministic system that’s already driving our cars, flying our planes and looking after our children - humans! Much like complex artificial machine learning systems - recognising there are still some differences in the way AI makes decisions - you can’t analyse the inner workings of someone’s brain and find out exactly why they made a certain decision. Instead you trust in their past performance and certified training, and put appropriate safeguards in place.
 
Developing trust in the performance of autonomous systems is a critical step before they can become more widespread. However, if we can build up the data and make it the data clear enough to prove they are safe and effective, I believe that people will accept that autonomy can improve our lives and make us safer.
There’s another non-deterministic system that’s already driving our cars, flying our planes and looking after our children - humans!

Dave Short

, Technology Director, BAE Systems
After all, globally well over a million people die every year in automobile accidents, caused by human drivers. I’m confident that autonomous systems will eventually help us reduce this number significantly.
 
To hear about other areas where autonomy can help us, including engineering, defence and medicine, please sign up online to watch the debate. We also talk about the ethics of autonomy, which is vital for every industry and organisation involved. In the defence industry, our position is that there are obvious benefits to unmanned systems, but that there should always be a ‘human in the loop’ when it comes to key decisions, including the use of lethal force.