Read the full paper here
Automation captured the imagination of the early Egyptians and Greek communities who built humanoid ‘Automata’, simple self-operating machines that follow a predetermined set of instructions. Modern automation is defined as ‘the use of machines and computers to operate without needing human control’. Modern machines do more than just reduce cost. Automated machining systems are highly accurate, so they also improve the quality of the final product. In the early days the benefits anticipated were that humans could enjoy leisure time while the robots worked.
Modern Artificial Intelligence (AI) has seen breakthrough after breakthrough. Development is accelerating, especially in fields such as machine and deep learning. AI can be defined as ‘the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings’. AI has a first adopter advantage, especially in the context of national defence and peace keeping. Early adopters have more time to develop systems and will have built up more data on which to train their systems.
The two robotic systems Ocado acquired in 2020 will save £7 million per year in each fulfilment centre. Amazon’s acquisition of Kiva Robotics, has led to a 22% fall in operating costs. One of the benefits of AI and automation is speed. In 2013 a new $300 million transatlantic fibre-optic link was installed saving 5.2 milliseconds, gaining high frequency traders billions of pounds. ‘Jigsaw’ is a tool used by the US Air Force to streamline the planning of in-flight refuelling. It is said that Jigsaw saved $12 million in fuel costs in the first month, approximately 2% of the entire DoD fuel budget. NASA currently use AI for scanning deep space and creating 3D models of asteroids to identify threats to the earth. The software uses Neural Networks to measure asteroid size, shape and spin rate. This AI based tool takes four days to categorise an object as threat or ‘non-threat’ compared with an average of four months for a human astronomer
One of the main differences between civilian and military systems is the environment. Military standard equipment has always needed to operate in a very broad range of temperatures. The weather conditions can vary substantially. A Canadian study identified that Unmanned Aerial Vehicles (UAVs) had the potential to save around 50% of the ‘per flight’ cost. However, they did not provide an all-weather capability and so the UAV could not fully replace the existing manned platform. Comparing military and civilian systems, there are a limited number of ‘unusual phenomena’ that could occur on a car journey, particularly on a motorway, as compared with ‘green lane’ driving. The extremely unpredictable nature of conflict means that it is much more difficult to design and test military systems. So civilian systems will be likely to move into the upper levels of automation before military systems. The amount of investment in a system will also impact on the level of automation that is possible.
There are risks involved in automating systems. Automated systems can behave unexpectedly, particularly in a complex environment where their training data is sparse. This will lead to concern about using such systems, which may hold back their development if this is not managed carefully. Although human operators also make mistakes.
Despite the risks, the use of Automation and AI have a place within military air and space systems. AI is becoming increasingly capable, potentially better than humans when it comes to long range identification because robots do not need to protect themselves and can be self-sacrificing if needed. The UK government has emphasised that weapons will always be under human control. There may be a time when it would be sensible to allow more of a nation’s lethal systems to make the final fire/don’t fire decisions, but still under the control of a human operator. Adversaries have always been good at spotting weakness in a military system and they will exploit this. In the commercial world, speed of response has become an important competitive advantage. This may be increasingly important in the military world in which both ‘no response’ and ‘the wrong response’ lead to escalating conflict.
Automation in the commercial world is progressing very quickly, although companies like Ocado are not at present investing in fleets of fully autonomous delivery vehicles in the near-term, to save the cost of a driver. This is a more complicated problem. In military systems the ‘art of the possible’ may be to rapidly draw on commercial technology and automate military systems to lower levels of automation than their commercial counterparts, going further over time when additional data and evidence about safety is accumulated. It will be important to consider the minimum viable system for each potential automation application and how this can be developed in parallel with existing commercial capability and how they could be used if they were really needed?
This piece is a precis to the paper An Introduction to Automation and Artificial Intelligence, written by Professor Martin Parr and published by the Freeman Air and Space Institute, King's College London.