Autonomous cars are classified in 6 categories, from Level 0 to Level 5.

Startup Profile AImotive: Global Accessibility to Self-Driving Cars

The idea of using artificial intelligence in the automobile industry is not new. However, fully self-driving cars or other autonomous vehicles are yet to come to our streets. To get there, there are many obstacles to overcome, from agreeing to a technological standard to finding enough qualified employees. One company that is already in full swing to bring autonomous cars to the next level is AImotive. We talked to Árpád Takács about challenges and opportunities in this area.

Thank you for sharing some insights about AImotive. In a nutshell: what do you do and what problem do you solve?

AImotive’s vision is to bring global accessibility to self-driving vehicles, regardless of driving conditions. We are developing the full ecosystem of self-driving, through AI-powered algorithms and technologies, which we are utilizing in our aiDrive, aiKit and aiWare suite of products.

In a nutshell about our products and functionalities:

AImotive has developed a full-stack software suite, aiDrive, for fully autonomous cars, providing a hardware-agnostic, scalable solution. Based on the idea that self-driving cars should mimic human behaviour, the algorithms rely on cameras as primary sensors for accomplishing the tasks of object recognition and classification, localization, decision making, trajectory planning and vehicle control. The software engine components are aided by an extensive toolkit, aiKit, to accelerate the training and verification, including calibration, data collection and augmented data generation, semi-supervised annotation, and a real-time, photorealistic simulation environment. In order to achieve a power-efficient implementation, AImotive is developing aiWare, an application-independent and universal AI-optimized hardware IP, which allows us to build hardware tailor-made to the exact needs of any given NN structures, including those used for autonomous driving.

How did you personally first come into contact with self-driving cars or smart cities?

I wrote my PhD dissertation on the topic of surgical robotics, thus I have been actively following the world of robotics for some years now, where self-driving cars are one of the hottest topics. Two years ago, I joined AImotive as an AI researcher, working on neural network architectures for image recognition for autonomous vehicles, and, as an outreach scientist, I have been involved with the majority of research topics ever since.

AImotive approaches the issue of autonomous cars from a global perspective and you have offices in Hungary, California and Finland. How important is this global cooperation for you?

Due to the nature of our products and development projects, becoming a global company is the only way to go. While research and development is done mainly in Hungary, we strive to have a sales and/or business development representative on all relevant global locations, such as Europe, US, Japan and China.

Finland has very appealing regulations for the testing of autonomous vehicles on public roads (our first testing licence was obtained in Finland), and allows an excellent testing location for extreme weather conditions. California or more precisely, the Silicon Valley is the home to all global technology companies, many self-driving startups and investors. Also, as we became the 36th company to obtain a testing license in California, it is an important location for technology development as well.

Autonomous cars are classified in 6 categories, from Level 0 to Level 5.

A Level 3 autonomous car, with a basic autopilot functionality for mid- and high-speed manoeuvres is the next milestone of AImotive.

What do you consider the biggest success for AImotive so far? What milestones do you want to reach next?

This year, AImotive has been chosen to be among the 100 leading AI startups in the world (CB Insights AI 100), and our global presence on conferences, expos and in the media as invited participants is a great indicator that what we do is important, unique and acknowledged.

Some of our greatest achievements are our test licenses for Finland, California and Hungary, as these countries have different requirements for business potentials, legal/technological concepts and available platforms. AImotive satisfied all of these requirements.

On the technological front, we have built an ecosystem of self-driving car development, where all components (algorithms, tools, hardware) are offering state-of-the-art performance and quality, creating a strong fundament for the integrated development.

Our next milestone is to create a Level 3, basic autopilot functionality for mid- and high-speed manoeuvres, extending our global presence to new locations, and obtaining new test licences around the world.

Your goal is to make self-driving vehicles accessible globally. What do you see as the biggest obstacles to that at the moment?

There are two large ‘schools’ of self driving: the LIDAR-first, and the camera-first approaches.

The former requires a detailed 3D, updated map of the world, which is very difficult to record and maintain, while expensive LIDAR technologies are limiting their global access in production. AImotive is using a camera-first approach, which does not require these 3D maps, and offer a better scalability. However, highly detailed (HD) maps are still an essential tool for self-driving, which are – although containing much less data and having better robustness – still challenging to collect and maintain.

Another obstacle (if not the biggest) is the different environment, weather and infrastructure conditions worldwide. Neural networks need vast amounts of data for training, and collecting a diverse, general data set for training from all over the world is a challenge as of today.

What are short-term and longer-term opportunities for self-driving vehicles in Europe?

The short-term opportunities mostly lay in cargo and public transportation, where Level 3 self-driving functionalities are already offering a safer and optimized way of transportation, even if only on pre-defined roads.

In the long-term, self-driving vehicles will shape the mobility and structure of cities in Europe: as our continent is relying less on cars than the USA, the integration of self-driving vehicles to existing public and mass transportation networks is a great opportunity for an even better mobility.

There are two approaches to autonomous cars: the LIDAR-first or the the camera-first approach.

The algorithms in AImotives self-driving cars rely on cameras as primary sensors for object recognition, classification, localization, decision making, trajectory planning and vehicle control.

Is there a specific development you are excited about currently? And, on the other hand, do you think there are specific obstacles that need to be overcome to achieve truly smart cities?

Self-driving technology is evolving rapidly, with new ideas and opportunities coming up every week. Every development is interesting in some way, of course, and we are primarily looking at the possibilities of their integration to our system. For example, V2X communication, map crowdsourcing, and solid-state LIDARs are technologies under development, and self-driving car developers need to be open for integrating these as they are ready.

What role do you think startups will play in changing the status quo and implementing smart cities?

Startups have always been referred to as agile, flexible and innovative companies, regardless of the sector or topic. The automotive domain is dominated by traditional, multinational companies with great experience in productization, testing and safety integration. Inherently, these tasks take a long time for development and deployment, following strict development standards.

The faster and innovative viewpoint of startups can help these giants with prototyping, experimenting with new technologies, speeding up the deployment of self-driving technologies and making cities smarter.

How do you judge the overall career opportunities in the smart city sector?

The smart city sector is dominated by highly qualified IT experts, software engineers and computer scientists. Globally, there is a high demand for these recruits, but the lack of qualified job seekers on the market is a problem everywhere.

Thus, career opportunities are great and many, which is good news for those who are looking for a job.

AI-powered algorithms and technologies in self-driving cars need to track and predict surrounding motion pattens.

The ability to track and predict the motion patterns of surrounding objects is a key feature in defining the safe and drivable space for autonomous cars.

What characteristics and skill sets should someone bring to the table to have an impact? What traits would you deem most important to successfully move into the smart city sector as a university graduate, startup entrepreneur, or job seeker?

University graduates and job seekers – if applying for tech positions – need:

  • good programming skills
  • excellent mathematical skills
  • the ability to work independently and change tasks in a flexible way

It is always an advantage if they have experience in what they would like to do, whether it comes from a previous job or a home-project.

Entrepreneurs and new, innovative companies are always welcome in the sector. One of the most important aspects here is that no matter how innovative the idea is, it is crucial to understand the way to productization through traditional standards. Smart cities, especially self-driving cars, are safety-critical systems, where each product has to comply to the highest standards in terms of safety.

What advice would you give graduates who want to join the smart city industry or a smart city startup?

The development and research topics related to smart cities are something that you don’t learn at school—take this opportunity to learn from others, colleagues, and enjoy the challenge.

Which three qualities should your ideal candidate have?

  • Excellent programming skills (C++)
  • Excellent mathematical skills and analytical thinking
  • Experience with automotive development

Árpád, thank you very much for this interview.

About the interviewee

Árpád Takács, Outreach Scientist at AImotive

Árpád Takács received his mechatronics and mechanical engineering modelling degree from the Budapest University of Technology and Economics. Before joining AImotive, he worked as research assistant at the Antal Bejczy Center for Intelligent Robotics at Óbuda University, and was part of the R&D team of the Austrian Center for Medical Innovation and Technology.

His fields of expertise are analytical mechanics, control engineering, surgical robotics and machine learning. Currently, he is an AI researcher and outreach scientist at AImotive, leading technology communication and public relations, promoting the self-driving car technology to the public, and representing the company at strategic discussions with regional decision makers and governmental institutions.

Contact

Hungary office
Szépvölgyi út. 18-22.
1025 Budapest, Hungary
Email: [email protected]
Web: aimotive.com
Social Media: Facebook | Twitter | LinkedIn

AImotive develops AI-powered algorithms and technologies for autonomous vehicles


Looking for more interviews with smart city pioneers?

  • Learn about CleverShuttle’s vision of an innovative driving service, that is eco-friendly and low-priced and tailored to urban areas.

For more insights into autonomous vehicles, check out our in-depth feature on self-driving cars.


Don’t miss upcoming news or job openings. Simply subscribe to our newsletter or follow us on Facebook and Twitter!

Theresa Kern
Follow us

Theresa Kern

Business Development at beta|careers
Theresa is a management student and team member of the 2017 class at beta|careers.

Her responsibilities include growing and nurturing our platforms and, occasionally, taking care of bus tickets for the team. In her free time, she can be found on horseback out in the woods.
Theresa Kern
Follow us

Hungary, Interviews, Startups