All the declinations of artificial intelligence

The evolution of robotics has led in recent times to an indissoluble link between robots and artificial intelligence, with the aim of improving factory functions and being an enabling element for the process of digital transformation. Marco Filippis, Product Manager Robot South EMEA & Export Marketing Coordinator at Mitsubishi Electric Europe, spoke about this during the Robotic Days in March.

Let us pretend that there is still someone who does not know you; I would therefore start by asking you to speak briefly about the construction philosophy and technological fundamentals with which Mitsubishi Electric approaches the world of robotics, and if you could provide an outline of your range of robots?

Mitsubishi Electric is one of the three major global players in the world of industrial automation, with a long history behind it. Indeed, this year the company is celebrating 100 years of history, marked by technological innovation. In the product portfolio, robots play a decidedly fundamental and strategic role, which envisages continuous evolution based on three conceptual pillars: Intelligence, Integration and Safety.

In the product portfolio, robots are taking on an increasingly fundamental and strategic role.
In the product portfolio, robots are taking on an increasingly fundamental and strategic role.

We live in the era of Digital Manufacturing and all that it implies in terms of technological, cultural and social changes. The evolution of robotics, especially, has recently led to an increasingly indissoluble link between robots and artificial intelligence. In this domain, Mitsubishi Electric seems to be accomplishing great results, could you describe them?

The evolution of robotics has recently led to an increasingly indissoluble link between robots and artificial intelligence, with the aim of improving factory functions and being an enabler for the digital transformation process. In this respect, Mitsubishi Electric has launched a milestone linking its entire product range to a philosophy called MAISART, an acronym for “Mitsubishi Electric’s Artificial Intelligence creates the State of ART in Technology”. The application of MAISART to the world of robotics envisages different real-life applications which have as their common denominator both the interconnection, horizontal on the shopfloor and vertical towards IT systems, and the sharing of the workspace with operators in complete safety.

On a purely practical level, could you provide some concrete examples of the combination of AI and robotics?

At a purely practical level, the combination of AI and robotics can certainly be found in different examples. Robot Motion Planning is a feature of Mitsubishi Electric robots, which, thanks to artificial intelligence algorithms and the aid of telescopic cameras, enables the creation of a completely destructured environment where the robot is able to adapt autonomously to variations in the surrounding environment. In this case, the presence of operators and their intervention on the parts to be manipulated are such that the environment becomes collaborative but at the same time the robot modifies its trajectory autonomously in real time. The Optimized Learning function involves the use of an AI-controlled force sensor on the robot wrist which allows the speed profile to be optimised during the assembly phases. Intelligent management of the speed profile increases the productivity of the line in complex assembly steps with small insertion tolerances, speeding up the traditional operation by 40%. An example is the assembly of high-density connectors/watches where accuracy is in the hundredths range.

The evolution of robotics has recently led to an indissoluble link between robots and artificial intelligence.
The evolution of robotics has recently led to an indissoluble link between robots and artificial intelligence.

Artificial intelligence is often associated with the concept of predictive maintenance of machine tools. Does the same apply to a robot cell whichever it may be?

Predictive Maintenance is one of the fundamental aspects of the application of AI to robotics inasmuch as it is possible to avoid downtime thanks to the comparison between the real robot and its Digital Twin, which inherits all the functionalities of the real twin, enabling constant real-time monitoring of the robot’s parts in advance. Besides, it is possible to accompany the maintenance phase by making the operator wear augmented reality visors which guide him step by step, avoiding human errors in the maintenance phase.

There is a lot of talk about collaborative robots, but I cannot help thinking that an important benefit may be found in so-called collaborative applications which do not necessarily have to pass through the application of a cobot. Am I wrong?

You are absolutely right! The Industry 4.0 paradigm shift should definitely be credited with creating a flywheel effect for collaborative robotics, but it also brought with it an aspect which is not quite correct, namely that cobots can be used in any application to make it collaborative. In reality, however, one must consider different application aspects which very often show that collaborative robotics is not the best choice. For this reason, Mitsubishi Electric has implemented a complete collaborative approach which not only includes the use of traditional industrial solutions or purely collaborative solutions, but also the possibility, through the MELFA SAFE PLUS advanced safety module, of creating collaborative areas around an industrial robot. In this way, the application acquires the advantages of both solutions depending on the interaction time of the operator and the machine.

Robots are increasingly changing their skin to adapt to a “Human-Centric” factory concept.
Robots are increasingly changing their skin to adapt to a “Human-Centric” factory concept.

The objectives of the digital transition aim at improving productivity, reducing production costs, raising quality standards and reducing lead times. All very important, but am I wrong in saying that in reality everything has to revolve around human-machine interaction, in terms of safety and ergonomics?

Man-machine interaction has taken on a fundamental character, and the Digital Transformation is fostering an increasingly lean and flexible factory concept. It is precisely for this reason that robots are increasingly changing their skin to adapt to a “Human-Centric” factory concept in which man is central and the robot takes its place according to the conceptual model for which it was designed. This phase is undoubtedly a delicate one, as a transition has to be managed between the traditional factory and an innovative model in which robots and humans share tasks and work spaces, but it can probably be seen as the most emblematic expression of the fact that robots do not steal work, but facilitate it.

Digital Transformation is encouraging a leaner and more flexible factory concept.
Digital Transformation is encouraging a leaner and more flexible factory concept.

In 2001, Mitsubishi Electric pioneered the collaborative factory represented by the “e-F@ctory” concept, which you have been experimenting with and implementing in your Japanese factories ever since. So I would like to ask you, finally, to look to the future and give us your definition of Factory 5.0 and tell us about the role that robots will play within it? Where is research taking you?

e-F@ctory is an innovative factory idea which was defined by Mitsubishi Electric in 2001, some 11 years before the Industry 4.0 paradigm. This certainly denotes a focus on the technological aspects of the factory, but above all the long-term vision of Mitsubishi Electric. Today, just like then, the challenges are certainly valid and stimulating, and in the world of robotics at Mitsubishi Electric there is the same inclination to innovate, even anticipating market trends. Suffice it to say that in terms of robot maintenance, we are in an advanced experimental phase which involves the use of augmented reality visors where the various maintenance steps are free from human error. Another particularly fitting example is the use of 5G infrastructure for remote applications, training or maintenance. Let us imagine having a cobot in Milan and one in Tokyo: thanks to the very low latency offered by 5G we are able to manually move the cobot in Milan, and in real time the cobot in Tokyo performs exactly the same movements without the difficulty of having to program the robot.