Search

Forecasting pedestrian trajectories in dynamic scenes remains a critical problem in various applications, such as autonomous driving and socially aware robots. Such forecasting is challenging due to human-human and human-object interactions and future uncertainties caused by human randomness. Generative model-based methods handle future uncertainties by sampling a latent variable. However, few studies explored the generation of the latent variable. In this work, we propose the Trajectory Predictor with Pseudo Oracle (TPPO), which is a generative model-based trajectory predictor. The first pseudo oracle is pedestrians' moving directions, and the second one is the latent variable estimated from ground truth trajectories. A social attention module is used to aggregate neighbors' interactions based on the correlation between pedestrians' moving directions and future trajectories. This correlation is inspired by the fact that pedestrians' future trajectories are often influenced by pedestrians in front. A latent variable predictor is proposed to estimate latent variable distributions from observed and ground-truth trajectories. Moreover, the gap between these two distributions is minimized during training. Therefore, the latent variable predictor can estimate the latent variable from observed trajectories to approximate that estimated from ground-truth trajectories. We compare the performance of TPPO with related methods on several public datasets. Results demonstrate that TPPO outperforms state-of-the-art methods with low average and final displacement errors. The ablation study shows that the prediction performance will not dramatically decrease as sampling times decline during tests., 14 pages, 7 figures, 2 tables. arXiv admin note: substantial text overlap with arXiv:2002.00391. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice

Forecasting human trajectories is critical for different applications, such as autonomous driving and social robot. Recent works predict future trajectories by using a generative model, in which human motion is encoded with recurrent neural network. However, the latent variable needed in the generative model is always either a random Gaussian noise or encoded from a scene. In this work, we focus on generating the latent variable from the trajectory itself. Specifically, we propose a latent variable predictor, which can bridge the gap between latent variable distributions of observed and ground truth trajectories. We evaluate the proposed method on several benchmarking datasets. Results demonstrate that the proposed method outperforms state-of-the-art methods in average and final displacement errors. In addition, the ablation study indicates that the prediction performance will not dramatically decrease as sampling times decline during tests.