The case for human-robot team-working in the medium- to high-value flexible manufacturing sectors now seems clear. In fact, although this is a sensible set of application domains to start with, in the fullness of time, the developed ideas could also penetrate lower-value, higher-volume, sectors if successful.
There is a valuable trade-off to be made here between the requirements for essential, but typically low-volume high-dexterity working, with the benefits from higher-volume automation that also typically brings with it a more easily regularised level of quality control. Finding the ‘sweet-spot’ for this trade-off is driving research and development towards conceptualising, and then instantiating, some form of ‘confederation’ between human and robot workers, epitomised by the notion of ‘HRI-teamwork’.
However, in any close-proximity interaction between a human being and a physically instantiated autonomous system that is powerful enough to be useful, considerations of the efficiency of the collaboration between human and machine and, above all, the safety of the human in this setting, quickly become critical. To be clear, although one must always first consider the more formal validated safety of the human, but also the wider factors of the contextual appropriateness of the robot’s behaviour.
With respect to the latter aspect above, this appropriateness of behaviour is critical to the development of that tricky human concept of ‘trust’. In other words, ‘trust’ can then lead to an efficient ‘teamwork’ approach to the achievement of what should be, from the human worker’s perspective, common shared goals that need to be fulfilled together.
However, trust is a tricky concept. We all feel that we know what it means when we perceive it in others or feel it ourselves, but there are plenty of cases where trust is allocated to a machine at a level that it has not really demonstrated to be deserving of or, conversely, trust is not felt when the machine has only acted correctly at all times.
Such developments, therefore, will require the encapsulation of both significant extension of existing Verification and Validation techniques into the prodigiously complex settings of HRI so as to ensure safety, but also an integration between this work and the cognitive-AI required to foster and sustain some kind of teamwork ‘feel’ that tries to encapsulates the concepts of attracting a level of trust that is demonstrably deserving.