Innovative developments

To achieve the overarching goal of safe, easily scalable traffic automation through decentralised intelligent vehicles and shared experience, three innovations are being developed in DEKOR-X:
​
Vehicle-based decentralized intelligence
​
This project is intended to demonstrate that data can be directly exchanged between vehicles. Intelligent infrastructure (e.g. sensors on traffic lights) remain expressly optional. One hypothesis is that while the decentralised approach will increase the demands placed on the algorithms, in the end, the “natural” scalability of the system will make this solution more robust and cost effective but still allow it for real time implementation. Because data in the decentralised architecture is only collected, processed and stored in the immediate spatio-temporal environment of the vehicles, the recommendations of the Ethics Commission on Automated and Connected Driving for needs-based, data-minimising information analysis can also be implemented.
​
Back-end models for delay-uncritical functions
​
Not all traffic information must be relayed to intelligent vehicles in real time. For certain functions, a certain latency (delay) is acceptable, such as the optimal, socially accepted driving trajectory within a lane or the shape and size of dangerous areas. Features unaffected by delay should be reproduced in the back- end with the help of mathematical models. From there, this model information can be quickly provided to the vehicles to support optimal traffic behaviour in terms of efficiency, acceptance and emission reduction.
​
Practical testing of various junction scenarios
​
In DEKOR-X, various applications are tested in real-life situations, research is conducted on how they are combined and then integrated into demonstrators as examples. To ensure interoperability, a closed test-junction is used so that all partners can test their developments with their vehicles in every phase of the project. The following applications are planned:
​
​
“Looking around corners” for increased safety
Every connected vehicle will use standard protocols to share its perception of the surroundings with other vehicles around them in real time. All vehicles independently generate congruent models of the environment and react to the traffic situation with autonomy. This enables each vehicle to circumvent view-blocking elements such as buildings, and proactively see “around corners” using the sensors of other vehicles.

“Optimum traffic behaviour” for efficiency, acceptance and emission reduction
The majority of manually driven vehicles move along optimal driving trajectories that are accepted by other road users. In this use case, these repeatedly used driving routes and tracks will be recorded by the vehicle swarm to form a delay-uncritical spatio-temporal model. This back-end model can then be used by automated vehicles to autonomously select the optimum, commonly accepted driving trajectory.

“Pedestrian crossings with experience” for a continuous learning approach to safety
Pedestrian crossings or other areas that require vehicles to pay particular attention are also stable in the long term. Vehicles can constantly record the spatial extent of them and merge this data to create a model in the back-end. When conveyed to automated vehicles, this model helps them predict dangerous situations and select the appropriate driving strategy.
