Diffusion method improves goal recognition from partial movement data
WoRD: Diffusion-Based Posterior Inference for Online Goal Recognition
Artificial IntelligenceMachine LearningRobotics
Summary
Knowing what someone is trying to do based on their movements is tricky, especially early on when their path could mean several things. The authors introduce WoRD, a technique that learns how agents move toward goals and updates its guesses as it sees more. This method keeps multiple possible goals in mind without getting overconfident, even when the observed path is unclear. WoRD was tested on robots, pedestrians, and vehicles, showing better and more balanced predictions than previous methods.
What this means in practice
- •For robotic system developers: Improve robot understanding of human or agent intentions early in their movements for better interaction.
- •For autonomous vehicle engineers: Enhance prediction of pedestrian and vehicle goals from partial trajectory data to improve navigation safety.
Authors
Almog Anschel, Sarah Keren
Abstract
Goal recognition is the task of inferring an agent's intended goal from its observed behavior. Most existing work focuses on high-level symbolic recognition, abstracting away the continuous motion through which agents reach their goal. Methods that do reason over continuous trajectories rely on an explicit dynamics model that is hard to acquire and rarely transfers across settings. Neither approach explicitly learns how agents move toward each candidate goal, limiting accuracy on ambiguous trajectory prefixes where such knowledge is most needed. We present WoRD (Waypoint Recognition with Diffusion), which uses a learned trajectory prior to infer the agent's goal from partial observations. At each step, WoRD runs one warm-started denoising chain per candidate goal, steers each chain toward the observed trajectory through a closed-form log-likelihood gradient, and reweights the resulting samples by Bayes' rule into a calibrated posterior over goals. We evaluate WoRD on simulated robot settings, pedestrian trajectories, and vehicle trajectories using four metrics targeting distinct failure modes. Across all three settings, the same inference procedure preserves calibrated multimodal beliefs through ambiguous prefixes where filter- and classifier-based observers collapse to a single mode or grow overconfident.