Mission: unmanned mobile machines. Why do we need bulldozers mimicking "The Terminator"?
Autonomous or automated mobile machines open up a world of possibilities to improve energy efficiency and profitability in a number of industries. We need to eliminate humans from the vehicle's cabin – the sooner, the better.
Heavy mobile machines have been a challenge for the developers of automation, and the journey towards autonomous machines has not been as quick as the development of artificial intelligence (AI) may have led to believe. Heavy machinery in, for instance, the forest or mining industries typically operate in demanding off-road conditions that the machine further alters.
At present, the automatic control of mobile machines is based on a combination of real-time simulation, the feedback of measurements by the machine, and pre-programmed control practices. For mobile machines of the future, this is by no means sufficient. To optimise the overall efficiency of a machine by having it focus purely on the task it is meant to perform and to maximise its energy efficiency, the machine's automation level must be raised considerably.
An anticipating machine aware of itself and its surroundings
Combining artificial intelligence and high-power computing enables creating a new way to control machines. High-power computing uses simulation that is significantly faster than real time. This makes it possible to predict behaviour in a large number of motion scenarios. Simulations up to 500 times faster than real time help to analyse a set of possible functions and select the optimal one. The most appropriate function and motion are chosen with the help of AI. As a result, the machines can be equipped with an awareness that enables them to understand the cause-and-effect relationships of their motions and operation.
"The machine houses embedded simulation models comparable to electronic stability control in passenger cars. The significant difference is that extremely efficient computation is used to give the machine the ability to predict future events and choose its movements optimally," explains Aki Mikkola, Professor of Virtual Design.
However, accurate and detailed simulation and self-adaptive control systems based on AI are not yet part of today's machine control. Similar methods have been successfully implemented in the gaming industry, but Aki Mikkola and Perttu Hämäläinen, Professor of Computer Games at Aalto University, are now attempting to introduce them into the control of heavy mobile machines.
Mikkola says that the collaborative project will employ Professor Hämäläinen's AI-based movement algorithm in the development of a machine's autonomous movements. To this end, simulation based on game engines is replaced with physics-based simulation, which can predict the operation of advanced machines accurately and computationally efficiently. In practice, physics-based simulation is performed with faster-than-real-time multibody dynamics. Mikkola himself is one of the most distinguished multibody dynamics researchers in the world.
Piloting an automated excavator
Continuously updating the technology of mobile machines is important for Finland's national competitiveness. Mikkola estimates that the manufacture of mobile machinery amounts to roughly a quarter of the entire Finnish metal industry production.
"Our results can help many industries. Our corporate partner network represents the entire spectrum of mobile machinery."
Mikkola states that the so-called flagship of the collaborative project is the virtual demonstration of an automated excavator that makes independent decisions just like "The Terminator". The autonomous excavator can adapt to a changing environment. Its virtual testing is carried out in LUT's SIM studio in Lappeenranta.
In many fields of industry, the next stage in the evolution towards fully automated machines will likely be remote-controlled machines. This will already have significant advantages.
"When a human being is no longer needed in the cabin, the machines can be designed purely from the standpoint of functionality. The physical dimensions, total weight, fuel consumption and emissions may drop radically."
LUT's and Aalto University's joint research project "Predictive simulation and control for mobile machines", headed by Aki Mikkola, received 230 000 euros from the Future Makers programme. The programme is jointly funded by the Technology Industries of Finland Centennial Foundation and the Jane and Aatos Erkko Foundation. It has provided a total of 3.2 million euros of funding for seven research projects that solve issues affecting the future of humanity.