“Nervtech is a high tech company. We offer services in the field of traffic safety, driver evaluation and training. Our goal is basically to approach the European goal of zero casualties on the roads by 2050. We are using driver evaluation and training by using our own technology, which is an advanced driving simulator, together with a number of additional software modules. We do this for insurance companies, so they can receive some information about their future drivers before they insure them, allowing them to get a basic insight into their behaviour in order to adjust premiums accordingly. In addition, we use our technology for different fleets of drivers, to address their training period, for example before they get hired, or for existing drivers to improve and/or to know what they are doing wrong. This can be related to driving safety as well as to the ecology of driving, such as reducing fuel consumption and the generation of emissions,” said Jaka Sodnik from Nervtech at May’s Innotalks at ABC Accelerator talking about this successful startup company.
“Originally our CEO Matej Vengust started the company almost five years ago. I approached him because I am a professor at the Faculty of electrical engineering University of Ljubljana and we were using driving simulations for different studies related to the human factor in driving. I approached Matej because he was advertising that he had this amazing simulator designed for a variety of purposes. Initially the company was just a two-man band who put together a small racing simulator. We then started discussing the possibilities of this simulator and that is basically how it all started,” said Sodnik, recalling the beginning of Nervtech’s startup story.
“First the simulator was used as a tool for racing entertainment to enable users to experience the feeling of being on a real road on a real track. This is achieved through an advanced motion system, which is the main part of our simulator that provides tactile feedback. When I approached the company, we started to focus more on the driver and human factor, such as how can people become better drivers, what can we learn from drivers, why drivers do different things in different situations, and so on. We started focusing more on serious areas such as traffic simulations, different environments and critical situations. Today we have a product that is already being used for driver evaluation and training, however, there are still many ways it can be improved. We are able to use the data about drivers’ responses to improve the system, and to learn – and this is also how machine learning is conducted.”
ABC Accelerator led Nervtech to the biggest insurance company in the region
In terms of the experience of working with ABC Accelerator, which Nervtech approached for help, Sodnik said that since he is a technical guy, scientist and engineer, he had no idea how to conduct business. I had fancy technology but I learnt that technology doesn’t have much in common with the product in the end, since, if you have something fancy but no one else thinks it’s fancy or understands how it will make their life better, then it’s a ‘non-starter’ – this was basically the most important lesson.
“The best experience was to get to know the mentors at ABC, to talk to them, since they have a completely different approach to the problem. They had no interest in the technical information, they just wanted to know, for example, how this product will make life better for insurance companies. Basically, it was invaluable for me to learn how to conduct business. We got our investors and connections through the ABC programme. It led us to a connection to the biggest insurance company in the region – Zavarovalnica Triglav. They would likely never have found us directly if we hadn’t gone through this programme. They saw our development here and the initial talks started at ABC. That was the first breakthrough.
Thereafter, we started working closely with Zavarovalnica Triglav, because, in addition to investing financially in the company, they also provided the know-how. They provided experts who supplied us with information about what the insurance company needs, their problems, and how to get more information about drivers from their perspective.”
Still in search of a sustainable business model
We still don’t have a completely sustainable business model for the business side of our company Nervtech. We are still making switches and accommodating changes, etc. Our initial idea was okay – we had a great simulator that can provide us with information about drivers, but now we needed to sell this information. However, we then discovered that even having sold five or six units to an insurance company, they encountered significant problems actually using it. The technology was too complicated, since an engineer is needed for it to be used. Then we changed the product into a service provider model. So now we own the technology but we offer services. Right now we are also changing the product to enable us to scale it.
From Triglav to American investor and back to another Slovenian company – AMZS
Nervtech currently employs around 20 people. Half of them are engineers, with a background in computer science, as well as experience of machine learning and artificial intelligence. The company now has a database of around 5,000-6,000 drivers in different situations, and Stanford University is analysing this data, making conclusions and recording patterns about driving. Following the link up with the insurance company Zavarovalnica Triglav, Nervtech acquired an American investor – Vector EQ – which comes from the automotive domain and can open markets up to us in that part of the world. The latest investor is AMZS, the Slovenian national traffic association. It is the main player in traffic safety in Slovenia and it was a strategic investment through which we got access to automotive clubs, explained Sodnik.
“Our long-term potential is to generate value from the data gathered. Right now we have specific drivers who are being evaluated and we give them scores, and we then pass this data to insurance and other companies. This deepens the level of interest. We also use the data to learn human behaviours. We make digital clones – replicas of drivers that are based on the data gathered – and can adjust clones to a drivers’ specifics, especially in terms of their driving behaviour. We can apply this technology to autonomous vehicles, so that cars can imitate people’s driving. A third-party provider, Volkswagen for example, can put new driving algorithms for autonomous driving into our simulation and this algorithm will encounter these clones that are based on drivers behaviour, so we can see how successful this driving algorithm would be in a real environment. If we have a simulation that is based on data about drivers in Ljubljana, this algorithm knows what to expect and when they will drive in Ljubljana,” continued Sodnik at Innotalks.
About autonomous driving…
“The problems of autonomous driving are basically infinite, because even if we are able to predict 99.99% of the situations that can happen, something new will always pop up, and the car, the system or the artificial intelligence (AI) will not know what to do in that case. All self-driving cars will ultimately be AI-based. However, these systems will need to ‘learn’, because it is impossible to pre-programme everything in the system for what will happen during the lifespan of a car. In unpredictable situations cars will ask drivers for help, and when the problem is solved the system will learn from that and next time know what to do. Cars will exchange information with each other all the time, thus they will constantly be ‘learning’. In terms of the infrastructure for autonomous driving, it will need to be greatly adjusted and accommodated,” concluded Sodnik.