Human Motion Prediction under Unexpected Perturbation
This addresses the challenge of predicting reactive and interactive human motions in uncontrolled scenarios, which is incremental as it extends motion prediction to a new, more complex task.
The paper tackles the problem of predicting human motions under unexpected physical perturbations, such as impacts involving multiple people, by proposing a Latent Differential Physics (LDP) model that combines differential physics and deep neural networks. The result shows LDP outperforms 11 adapted baselines, improving prediction accuracy by up to 70% with high data efficiency and strong generalization.
We investigate a new task in human motion prediction, which is predicting motions under unexpected physical perturbation potentially involving multiple people. Compared with existing research, this task involves predicting less controlled, unpremeditated and pure reactive motions in response to external impact and how such motions can propagate through people. It brings new challenges such as data scarcity and predicting complex interactions. To this end, we propose a new method capitalizing differential physics and deep neural networks, leading to an explicit Latent Differential Physics (LDP) model. Through experiments, we demonstrate that LDP has high data efficiency, outstanding prediction accuracy, strong generalizability and good explainability. Since there is no similar research, a comprehensive comparison with 11 adapted baselines from several relevant domains is conducted, showing LDP outperforming existing research both quantitatively and qualitatively, improving prediction accuracy by as much as 70%, and demonstrating significantly stronger generalization.