R2-D2: a System to Support Probabilistic Path Prediction in Dynamic Environments via "Semi-Lazy" Learning

Path prediction is presently an important area of research with a wide range of applications. However, most of the existing path prediction solutions are based on eager learning methods which commit to a model or a set of patterns extracted from historical trajectories. Such methods do not perform very well in dynamic environments where the objects’ trajectories are affected by many irregular factors which are not captured by pre-defined models or patterns. In this demonstration, we present the “R2-D2” system that supports probabilistic path prediction in dynamic environments. The core of our system is a “semi-lazy” learning approach to probabilistic path prediction which builds a prediction model on the fly using historical trajectories that are selected dynamically based on the trajectories of target objects. Our “R2-D2” system has a visual interface that shows how our path prediction algorithm works on several real-world datasets. It also allows us to experiment with various parameter settings.