Intelligent and self-learning traffic control with 3D & AI

Important results from the project

The project has, through live-tests, shown that 3D&AI-based sensors are a viable solution for traffic control in signalized intersections. The solution not only includes motorized vehicles but also cyclists and pedestrians in a natural way. The project also shown that wireless transmission signals between sensors and the signal controller with low enough latency is possible. The project is a collaboration between Viscando, Swarco and City of Uppsala.

Expected long term effects

The results show that the detection performance is as good as induction loops but with less risk of crosstalk between adjacent lanes. Detection of cyclists and pedestrians naturally provides the opportunity to increase the mobility of these road users in signalized intersections. The project has also developed a low-latency solution for wireless communication between sensor and controller, we expect that it will significantly reduce installation costs and thus increase the solution´s attractiveness.

Approach and implementation

Planning and preparation has primarily taken place via online meetings (pandemic restrictions). Three 3D & AI sensors were installed for good area coverage of an intersection in central Uppsala with relatively high flows of vehicles, cyclists and pedestrians. Equipment was also installed for synchronized event logging on induction loops and 3D&AI sensors. A semi-automatic comparison between these events was performed. The corresponding evaluation for cyclists and pedestrians was done manually. The wireless communication was tested on-site, including latency measurement.