Multi-person motion prediction captures diverse possible futures
FlowMoP: Stochastic Multi-Person Motion Prediction
Computer Vision and Pattern RecognitionMachine Learning
Summary
Predicting how groups of people will move in the future is difficult because many different outcomes are possible depending on their intentions and interactions. The authors propose a new method that generates many possible future motions for multiple people at once, taking into account how they move and interact. Their approach uses advanced techniques to make these predictions smooth and more accurate. Tests show it works better than older methods on several motion datasets, including one capturing professional soccer players. This method can better model uncertainty and social influence in movement prediction.
What this means in practice
- •For robotics engineers: Create robots that predict and smoothly adapt to the diverse future movements of multiple people around them for safer interaction.
- •For sports analytics teams: Model players’ future movements in team sports to inform strategy based on likely motion scenarios influenced by social interactions.
Authors
Aadya Agrawal, Ho Kei Cheng, Alexander Schwing
Abstract
Predicting the future motion of multiple people is inherently stochastic: the same observed history admits many plausible continuations depending on intent, coordination, and social context. Yet the dominant paradigm in multi-person 3D motion prediction remains deterministic, with existing methods producing a fixed number of outputs per agent. In contrast, we present a conditional flow matching framework that jointly generates diverse future motions for multiple agents, modeling both global trajectories and root-relative poses. A B-spline reparameterization compresses the generative state space while enforcing temporal smoothness, and a dual-stream transformer conditions each agent's predictions on observed neighbor motion for interaction-aware forecasting. Our method achieves state-of-the-art performance on CMU-Mocap (UMPM) and 3DPW, outperforming deterministic baselines on joint position error, pose error, and final displacement error despite producing a full predictive distribution. We additionally establish the first motion prediction baseline on WorldPose, a large-scale professional soccer dataset, where social conditioning demonstrably concentrates predictive uncertainty along interaction-constrained directions.