A Framework for Hybrid Location Prediction via Decision Tree Classification

Location prediction has attracted massive attention from various robotics research fields such as human-robot collaboration, trajectory planning in human-robot interaction and etc. By far many location predictors are proposed while none of them is capable of stably outperforming others when facing different mobility patterns. In this paper, a framework for hybrid location prediction is proposed with the idea that different location predictor is selected intelligently by analysing personalized mobility patterns. Experiments conducted on realworld datasets show the effectiveness of the proposed framework compared with single predictor. Furthermore, evaluation and discussion are offered to obtain more insights into this topic. Finally, the conclusion is drawn to summarize the entire work.