Nordic IT services and software company Tieto has developed a new test solution to help improve the urban traffic safety of pedestrians utilising artificial intelligence (AI) and Internet of Things (IoT) technology.

In collaboration with the city of Tampere, Finland, the solution can automatically detect when a pedestrian is planning to cross the street at an intersection. 

“By enhancing existing traffic monitoring technology with artificial intelligence, we can better identify traffic accident risks,” explained Jari Torkkola, Program Director, Tieto Product Development Services.

“The system monitors the movements of vehicles and pedestrians, and recognises when a pedestrian intends to cross the street. Especially in areas of limited visibility, the system can help prevent accidents.”​

It will then send an alert that can be relayed to the automatic traffic signals to help reduce reduce pedestrian and vehicle accidents in urban intersections without the need for any identification.

It is hoped in the future this system can be used to help direct vehicles, improve traffic flow and act as a ‘building block for autonomous vehicles.​’

Pekka Stenman, Traffic Engineer, City of Tampere, elaborated: “We had identified the most common types of accidents between vehicles and pedestrians. Using them, we built an algorithm that can predict the movement of vehicles and pedestrians on the street.

“The new solution has many potential uses in addition to boosting traffic safety. We already receive information about vehicle traffic, but not very much about pedestrian traffic. 

“We want to see how people move, and perhaps construct heat maps of Tampere’s pedestrian flows to assist with traffic planning. Another interesting opportunity is introducing more intelligence to traffic lights by identifying and predicting people flows.”

The test operated so:

  • An intersection traffic camera feed was connected to a cloud-based AI system which monitors vehicles and pedestrians.
  • When the system’s algorithms detected a pedestrian is beginning to cross the street, it provided an alert.
  • This alert can then be used to notify other connected systems, such as automatic traffic signs.
  • Results reveal a 99% accuracy rate ‘under ideal conditions; at night accuracy reached 75%.

“This implementation also provided one piece in the puzzle of autonomous vehicle systems,” continues Torkkola.

“A critical question is how self-driving vehicles are able to recognize and avoid obstacles. This type of pedestrian recognition system could be an important element in the safety of autonomous vehicles in urban areas.”