Behavior Prediction Based on Obstacle Motion Patterns in Dynamically Changing Environments
Zhuo Chen, Daniel C. K. Ngai, N.H.C. Yung · 2008
This paper proposes a behavior prediction method for navigation application in dynamically changing environments, which predicts obstacle behaviors based on learned obstacle motion patterns (OMP) from observed obstacle motion trajectories. A multi-level prediction model is then proposed that predicts long-term or short-term obstacle behaviors. Simulation results show that it works well in a complex environment and the prediction is consistent with actual behaviors.