‘Humans and robots need to cooperate’ – BBJ – Budapest Business Journal

Mollia Ltd. is a Hungarian technology startup developing adaptive and teachable software for the kinetic control of humanoid robots. The key novelty in their product is a linguistic style learning system that transcribes human movement in a uniform system of geometric building blocks. The company envisages that in the future, robots will be available for the average household. Still, while they wait for this to become a reality, they are developing a game in which regular people can teach robots various movements in the virtual space. If the game becomes successful, it will result in a broad set of motions that will eventually be transferable to actual robots for such applications as expeditions to Mars.

We talkedwith Mollia deputy CEODniel Vincz about the importance of robotics.

Although innovation is becoming increasingly valuable in any industry, digitalization, automation, robotics, and industry 4.0 solutions are not as prevalent among micro and small enterprises and are likelier to play an important role for medium and large enterprises. Do you agree with this statement?

My short answer is: yes. On the other hand, however, Mollia is not an industrial solution. Back in the seventies, people did not think that every household would have computers one day. Our team believes that humanoid robots will go through a similar path and will become common household items. Our goal is to develop adaptive and teachable software for the kinetic control of humanoid robots. The most important novelty in our technology is a linguistic style learning system that transcribes human movement in a uniform system of geometrical building blocks. This innovation enables us to improve robots general mobility. It also allows robots to learn new forms of movement and perform better in novel tasks. As a result, in the future, even regular people will be able to interact with robots and train them in different skills without any IT training.

You want to create a system in which robots can learn alongside humans, but at the same time, when a robot encounters a problem, it will be able to solve it within its area of competence. Is that correct?

Just think about it. Humans were able to rise above animals by becoming the sole bipedal mammal. It is a very complex task to walk on two legs. This directly correlates with how our brains evolved over several million years. It allowed us to use our brain capacity for other things such as communication. The link between communication and humans becoming bipedal is deeply ingrained in humanity. Billions of people on the planet walk on two legs, yet we do not use this ability, this common knowledge, to teach robots to do the same. Recognizing this was our starting point when we began developing our technology.

We are currently operating with 36 degrees of freedom, which refers to the number of ways the robot can move its parts. Users can equip robots with freely determined movements characteristics, which will be saved in the total range of the characters kinetic activity. This means the robots autonomous movement may be influenced externally, which results in a kind of true cooperation between man and machine. This is similar to how a horse and its rider communicate with each other. In their natural state, horses usually do not jump over 150cm hurdles, so a human must cooperate with the animal and try to get it to complete the task because the animal has a will of its own. The horse must be gently guided to do what the rider wants it to do. The horse and the rider must understand and be comfortable with each other in order to perform such a task. This is basically what we are doing as well: we are creating an external interface through which the human and the robot are able to cooperate. Currently, this interface consists of a mouse and keyboard as this can be found in almost every household. The long-term goal, however, is to have a device worn on the head that senses brainwaves and use it to teach movements to robots. I know this sounds very utopian, but I believe we will achieve this within the foreseeable future.

Why are we developing humanoid robots?

Mostly because this way, robots will be able to use the same tools as humans and vice versa. Furthermore, humans are likelier to positively relate to humanoid robots than robots with other forms, as this is the form we are most used to. Just think about the reactions people had to the Boston Dynamics videos, where the boxes were knocked out of the robots hands multiple times.

Where is the robot being developed?

Currently, development takes place in a Bullet Physics based virtual environment. This is a physical environment where we try to simulate the Earths gravity. This has both advantages and disadvantages. On one hand, we receive accurate feedback from everything and we know where the robot steps and the force with which it steps. On the other hand, we arent able to test its movement in the real world, like we would be able to with a physical robot. We are continuously working on developing the latter.

Babu is able to adapt to certain novel characteristics such as a new body type, but can also learn tasks on its own such as walking in heels. Furthermore, it is capable of handling objects with different weights in physical space. This high-level adaptive capacity is essential to enable it to interact with users. The goal for the robot is to be able to follow real-time kinetic instructions via input devices (mouse, keyboard, VR, movement tracking devices, etc.) while keeping key elements of its own movement that are necessary for maintaining balance. Humans and robots must be able to cooperate. This is training, which means the process is slow but will produce exciting results. Furthermore, we would like to be able to transfer knowledge from a robot that a user has finished teaching, to another robot, which will allow the new robot to learn new movements from a pre-existing level of knowledge, and thus improve even more.

As Ive mentioned before, the long-term goal is for the robot to reach households where it will be able to complete tasks that are appropriate for its set of movements. Robots trained in this waymay also perform activities that could be dangerous to humans, such as mining and deep-sea welding. Basically, you have to look at the whole thing from a positive perspective. Robots shouldnt be feared.

What are your short-term plans?

Before we are able to use our technology to control real physical robots, we need to create an extensive set of movements. In order to achieve this, we are developing a video game in which you can teach the robot various moves in a virtual world. This is very exciting because we are able to involve a multitude of people in robotics development, which has yet to be done. Just think about it: currently, robotics development is very resource intensive and mostly only scientists with PhDs can take part in it, whereas there are nearly eight billion people on the planet who have kinetic skills that can be used to teach robots. The best analogy is perhaps Wikipedia, which allows anyone to share his or her knowledge with others. In essence, we are trying to build the Wikipedia of robot movement.

If we get to the point of having a sufficient database of movement sets in virtual space, then we will be able to transfer these movements to physical robots, which will be able to implement more complex movements. Such a robot could be used, for instance, during an expedition to Mars, where the robot would be able to establish a camp before humans arrive.

What are your target markets?

We are open towards both Hungarian and foreign video game developers. We had meetings at the university where the developers of Angry Birds started, and based on their feedback, they see great potential in Babu. We are also in negotiations with a Slovak video game development company regarding the shared development of a video game. We believe that in the long term, anyone can become interested in using our product, but at the moment, we are primarily focusing on the North American market. One reason for this is our investor, Vespucci Partners, with whom we plan on reaching the Series A investment circle, which we would use to finance the next seven years. We are primarily negotiating with American companies regarding this, as the robotics industry is powered by the US and China anyway.

How well could a Hungarian startup compete in these two markets?

Its a little too early to come to conclusions, but we havent really seen any companies that deal with anything similar. It would be a potential opportunity for us to cooperate with Tesla. This August, Elon Musk showcased the Tesla Bot, otherwise known as Optimus, which is a general-use humanoid robot developed by Tesla, Inc. with an exceptional hardware solution for which we could provide the software. This summer, I received a scholarship to Draper Universitys five-week summer program in Silicon Valley. The programs founder is Tim Draper, an iconic figure in the Silicon Valley venture capital scene, so an introduction to Elon Musk is not such a farfetched idea.

Mollia Intro Video from Mollia on Vimeo.

How do you think robots will feature in the lives of average Hungarians and what role will they have in the future?

At the moment, robots arent that widespread in Hungarian households; however, in industrial settings, robotics solutions are quite common. To approach this subject from a more realistic perspective, we are also cooperating with the Antal Bejczy Center for Intelligent Robotics of buda University, where in addition to the latest trends, we pay close attention to what we could incorporate from the robots movements there. The goal of the center is to help the scientific and technological development of robotics primarily for service applications. It is an independent platform which widely leans on the most important partner institutes and companies in both Europe and North America. We believe that technology and thus robotics will become an important part of the average Hungarians life in the future.

What does Mollias vision of the future look like?

Our most ambitious plan is to implement our algorithms in the physical world to control real humanoid robots. One of the most important potential technological changes during the next decade could be robotics achievements transferring from the showroom to our homes, thus becoming a part of everyday life. We want Babu to become an important part of this process. We arent solely focused on robotics, however. We have several exciting ideas about how our team at Mollia could come up with something new in the area of NFTs, esports and even HealthTech.

I must mention here that we are constantly looking for new domestic talent in mathematics, software development, design, marketing, and robotics, so if you feel you are up for it, please do not hesitate to contact us!

Link:

'Humans and robots need to cooperate' - BBJ - Budapest Business Journal