University of Houston

Understanding of human social behavior can provide useful information towards the design of accurate and efficient algorithms for video surveillance applications. One challenging problem in video surveillance or robotics is the design of human motion trajectory forecasting algorithms. This in particular is an interesting problem because the outcome is influenced by many factors, of which we believe that the geometry of the environment plays a significant role. In addressing this problem, we have built a model to estimate the occupancy behavior of humans based on the geometry and certain social norms. We have also developed a trajectory forecasting algorithm that understands this occupancy and leverages it for trajectory forecasting in previously unseen geometries (environments). We perform thorough experiments to quantify the error between our prediction model and the trajectories obtained from real world human subjects. Experimental data suggests a significant enhancement in the accuracy of trajectory forecasting by incorporating the occupancy behavior model. Our current work deals with indoor scenarios in static environments. We propose to extend our work to a general outdoor environment and also include dynamic scenarios where moving objects can be accounted for during forecasting.