• a novel segmentation-based approach to estimate intention of a mobile agent. • an efficient implementation of a new segmentation-based algorithm for continues or realtime intention recognition. • evaluation of the latency, stability, and accuracy of the proposed algorithm compared to other state-of-the-art online intention algorithms. This paper proposes and evaluates an efficient online algorithm for recognizing the movement intentions of mobile agents. As an agent reveals new movements, our algorithm continuously ranks various possible intentions, such as reaching a specific destination or avoiding a particular area, using a novel approach to semantic trajectory segmentation. We empirically validate the performance of our algorithm against state-of-the-art alternatives in the path planning using simulated movement data. The results show that our approach improves reliability and accuracy in identifying true intention in scenarios with an increased number of candidate destinations, or when a mobile agent takes less planned and predictable routes.
Hashem et al. (Sun,) studied this question.