New-Tech Europe | February 2019

ture a maximum amount of wind energy and also ‘disturb’ the flow of the wind to any neighboring wind turbines as little as possible. Finally, smart systems evolve towards ‘Autonomy’. When a product is capable of monitoring itself or carrying out an action – and making that action as optimal as possible – it can work autonomously. For instance, there is the iRobot vacuum cleaner robot, which is capable of cleaning all sorts of surfaces in the home, as well as detecting dirt, finding its way round furniture and avoiding tumbling down stairs. It also ‘stores’ details of the layout of a room in its memory for the next time and makes its own way back to its recharging station, where it announces its safe arrival with a triumphant sound signal! Smart systems can also be connected with each other so that they can carry out actions in tandem, learn from each other – and so on. An ex-ample of this is the idea of driverless cars and the road infrastructure working together so that if there is an accident somewhere, cars further away from the incident can be notified and the appropriate action tak-en. As we can see from these examples, we will gradually evolve towards systems that are capable of learning and taking decisions by them- selves. And equally gradually, we – humans – will hand over the moni- toring, control and optimization, partly or in full, to machines. This will happen faster in some sectors than in others – and there are various reasons for that. In the mining industry, for instance, Joy Global’s Longwall Mining System is used to dig underground virtually automati-cally, without any human input. Staff sitting in the control room above the ground keep a close eye on everything going on and

Fig 1: In mining, for example, fully autonomous robots are used to work under the ground, while everything is monitored closely from a control room above the ground so that humans can inter-vene when necessary, for example if repairs are needed. (Video from Joy Global / Komatsu Mining via https://hbr. org/video/3819456791001/smart-connected-mining)

the cobot is able to carry out certain actions safely close to and with the help of its human workmates. This means that tasks such as applying adhesive can be carried out much more precisely, consistently and al- ways in the same way. Meanwhile, the cobot’s human workmate is able to control and direct it using hand gestures. There are still many challenges to overcome in the area of communica- tion between robots and humans. For example, will a robot ever be ca-pable of identifying our intentions? Can a robot detect if we say some-thing in a fearful or more self-assured way? In which case, this can be important in certain situations. Or when we carry out an action, what does this say about our actual intentions? For instance, it is no easy task to get a driverless car to recognize whether a pedestrian intends to cross the road, or is simply standing at the side of the road. Typical-ly, as a pedestrian, we will try to make eye contact with the driver to indicate that we

only send en-gineers below ground if it becomes necessary. So, for the sake of peo-ple’s safety, mining has evolved to the most advanced stage of auton-omy – although people are still very much a crucial factor of operations. People and AI systems will become workmates Human-like AI, human-centric AI, human-in-the-loop AI – these are all terms to indicate that human beings are still very much central to the story. Robots and machines need to be made in such a way that people can understand them, are able to communicate with them and can work efficiently with them. That way, machines can carry out tasks on behalf of and for the benefit of humans. A good example of this is the ‘cobot’, or collaborative robot, developed to assist Audi production line workers in assembling cars. Whereas pre- viously these types of machines used to be placed in safety cages,

New-Tech Magazine Europe l 21