Robotics in defence

Why it matters

Robotics and autonomous systems (RAS) are already extensively used in many civil domains, especially since technological progress render them less expensive and bulky while, at the same time, more flexible and easier to interact with. 

As such, robots and multi-robots (or swarm) systems (MRS) also have the potential to play, in the foreseeable future, a disruptive role in military operations in the sense that they will allow to perform task that today are considered too risky, complex or even impossible for humans. 

But RAS not only have the potential to perform conventional dirty, dull and dangerous military tasks like surveillance and counter mining; they are also likely to change the way military operations are conducted in the future, and even to make it possible to envisage new type of missions.

The way ahead

The next big challenge for defence will be to bring these technologies from the lab to real operations where robots will have to co-exist and cooperate with humans. To have a usable and useful RAS, simply increasing the level of automation will not be sufficient. Achieving trusted autonomy will be essential.

Just as it is the case in human relationships, robots must earn the trust of their teammates and operators through proven reliability. Barriers to establishing trusted autonomy include those normally associated with ’standard’ human-human trust relationships. But there are additional barriers associated with human-machine trust: different ways of thinking (digital vs. analogue) and expression, low transparency and traceability (robots can’t explain their own decisions), low mutual understanding of goals. The further the level of autonomy will be pushed by technology, the more the following questions will need to be answered: How far can robots be ‘trusted’ to perform their allocated tasks without the need for human supervision? Which level of trustworthiness can be associated with a task performed by an autonomous system? This is even more challenging in MRS where agents must also ‘trust’ other agents. 

Trusted autonomy represents a complex ’socio-technological’ research area which is still at an early stage and which will require, in addition to the establishment of the relevant legal and ethical framework, significant research in the following fields: development of more performing, robust and reliable sensing and data acquisition;  human-machine communication and integration; and verification, validation and evaluation methods.