Through a distributed intelligence, intelligent vehicles, the intelligent edge (route) and the intelligent cloud can help to develop and test novel mobility solutions and applications, altogether then contribute to an innovative mobility ecosystem.
Distributed intelligence for the urban mobility of the future
Edges | Sensors
The Data-analysis Platform (DP), developed based on Machine-Learning solutions, was used for parking forecasting, traffic intensity prediction and air quality analysis. For optimal predictions, an automatic transformation of the data into a usable format is carried out, which explicitly represents the essential influencing factors of the prediction and allows the use of the data in machine learning algorithms.
Since the up-to-date data is always required, the data streams are processed using real-time time series analysis methods. In these machine learning techniques, models are generated based on the past data points and they are continuously adjusted with the help of the current data.
In order to achieve such an update of the models, the difference between the predictions and the actual values at the time is calculated and used in the process. The newly updated learning models are then used to predict future trends.
Another focus is on learning artificial intelligence and deciding which of the learned models work best to select and run it for the prediction.
Vehicle analysis, pedestrian activity analysis, detection of unexpected crossings