noun_Email_707352 noun_917542_cc Map point Play Untitled Retweet Group 3 Fill 1

The future of safe driving and traffic lies in AI

AI offers new possibilities for safer traffic. Before they become reality, however, lots of collaboration is needed.

Sami Järvinen / May 19, 2020

AI offers new possibilities for safer traffic. Before they become reality, however, collaboration is needed between software companies, research institutes, universities, startups, cities, and public entities.

Safety is one of the drivers behind digitalization and technology investments in industries. Safety can be understood in many ways. In traffic, some may consider safety a feeling, while others require visible proof of accuracy or how the system operates. All things considered, technological assistants for humans or vehicles are always a matter of quality and trust.  

Even if an AI solution were 100% accurate, proven to save hundreds of lives and prevent thousands of accidents every year, regulations demand bulletproof evidence of this before such a system would even be considered for real life use. After acceptance, the culture of thinking also needs to change, as there is still lot of fear associated with handing human responsibilities over to automation.   

Automotive software developers at TietoEVRY have supported many Automotive partners in their product development projects aiming to increase the safety of driving and traffic. For example, pedestrian safety monitoring and city utilization analysis projects are interesting steps towards safer traffic. Cities and other infrastructure operators play a significant role in developing transparent and integrated solutions for urban areas.  

Effectiveness – and the largest impact – are only reached when data and connectivity meet use-case needs. In rapid environments such as highways or urban crossing areas, things can change dramatically fast. Thus sensing, observation, analysis, and informing need to be almost real-time. Decisions must be made locally as there is no time for moving data to back-end systems for calculation.  

Edge computing powered with Artificial Intelligence could be a solution in these situations. Naturally, historical data and behaviour should be analysed and learned somewhere with scalable processing power, but ultimately the operation is local and concrete.  

When aiming to decrease traffic accidents, the ultimate goal is minimizing human decision making. We know that risks increase whenever humans influence automated processes. There already are examples of this in aviation and manufacturing; why not in traffic management?  

As trust increases, autonomous and highly automated traffic solutions will slowly become reality, step by step via Advanced Driver Assistance Systems and Autonomous Driving solutions. Computer vision solutions are currently are under development for image enhancement and object recognition. They will enable sensor fusion types of solutions, creating holistic understanding for example of the certain behaviour or objects, possible traffic violations, or predict the need for maintenance and other resources.  

Parallel to increasing safety in urban areas, costs can also be adjusted while bringing in more scalable technologies and solutions that can be retrofitted to for example older camera technology environments. When integrating the communication systems, powering with the needed amount of calculations, and scaling up with multiple data sources and computing power, the solution will create concrete safety applications for several parties such as infrastructure, network, care, safety, and automotive service and solution providers.   

When it comes to the different parties and technologies used, the value chains and networks in automotive and traffic domains are wide. It’s important to join forces with software companies, research institutes, universities, startups, cities, and public entities to work around real use-cases together with Investors such as manufacturers, network and service providers, and solution providers. To come up with solutions, the expected solution accuracy level needs to be clear to all parties. Systems that have operations based on visual observation should provide close to 100% accuracy, as there is no room for misunderstandings. Artificial Intelligence in different forms is one of the main technologies enabling these applications.  

Whenever we want something new to become normal, culture needs to be changed as well. This is what we want to promote via low risk, highly beneficial, joint proof of concept examples. Safer traffic and society will become a reality through great collaboration.  

Sami Järvinen’s thoughts were featured in the Finnish automotive magazine Tuulilasi’s April 2020 report: Artificial intelligence for traffic and cars. 

Play video
Sami Järvinen
TietoEVRY alumni
Share on Facebook Tweet Share on LinkedIn