Human motion trajectory prediction using the Social Force Model for real-time and low computational cost applications
This addresses trajectory prediction for applications like human-robot collaboration and self-driving vehicles, offering incremental improvements in accuracy and efficiency.
The paper tackles human motion trajectory prediction by proposing SoFGAN, a model combining GAN, Social Force Model, and CVAE to generate plausible trajectories and reduce collisions. It shows improved accuracy on UCY and BIWI datasets compared to state-of-the-art models and enables real-time, low-cost predictions without GPUs.
Human motion trajectory prediction is a very important functionality for human-robot collaboration, specifically in accompanying, guiding, or approaching tasks, but also in social robotics, self-driving vehicles, or security systems. In this paper, a novel trajectory prediction model, Social Force Generative Adversarial Network (SoFGAN), is proposed. SoFGAN uses a Generative Adversarial Network (GAN) and Social Force Model (SFM) to generate different plausible people trajectories reducing collisions in a scene. Furthermore, a Conditional Variational Autoencoder (CVAE) module is added to emphasize the destination learning. We show that our method is more accurate in making predictions in UCY or BIWI datasets than most of the current state-of-the-art models and also reduces collisions in comparison to other approaches. Through real-life experiments, we demonstrate that the model can be used in real-time without GPU's to perform good quality predictions with a low computational cost.