Long-Term Recurrent Predictive Model for Intent Prediction of Pedestrians via Inverse Reinforcement Learning

Anonymous

Abstract: Recently, the problem of intent and trajectory prediction of pedestrians in urban traffic environments has got some attention from the intelligent transportation research community. One of the main challenges that make this problem even harder is the uncertainty exists in the actions of pedestrians in urban traffic environments, as well as the difficulty in inferring their end goals. In this work, we are proposing a data-driven framework based on Inverse Reinforcement Learning (IRL) and the bidirectional recurrent neural network architecture (B-LSTM) for long term prediction of pedestrians' trajectories. In the proposed framework, we firstly learn a reward function of the urban traffic environment scene that capture the preference of the pedestrians with respect to the scene's physical contextual information. Then based on the learned features of this reward function along with past trajectories of pedestrians in the scene, we forecast a probability distribution over the pedestrians' future trajectories using B-LSTM model. We evaluated our framework on real-life datasets for agent behavior modeling in traffic environments and it has achieved an overall average displacement error of only 2.93 and 4.12 pixels over 2.0 secs and 3.0 secs ahead prediction horizons respectively. Additionally, we compared our framework against other baseline models based on sequence prediction models only and we have outperformed them with a lowest margin of average displacement error of more than 5 pixels.