PT cubed: Prototype-Guided Self-Supervised Pretraining for Generalizable mmWave Human Point Cloud Representation

arXiv:2610.38551 · cs.AI · Submitted 2026-10-28 · Read on arXiv

cs.AI

Submitted: 2026-10-28

Updated: 2026-10-28

Comments: 30 pages

License: http://creativecommons.org/licenses/by/4.0/

The gist: Multi-source millimeter-wave (mmWave) observations offer an opportunity to learn human representations reusable across sensing tasks.

Terminology

Abstract

Multi-source millimeter-wave (mmWave) observations offer an opportunity to learn human representations reusable across sensing tasks. However, sparse detections and multipath effects make these observations unstable, challenging existing point cloud pretraining methods. We propose PT cubed, a prototype-guided self-supervised pretraining framework for mmWave human point clouds. PT cubed represents local states in spatial regions as sparse compositions of shared prototypes and learns their temporal evolution through autoregressive prediction. Observation consistency further stabilizes representations under radar perturbations. After pretraining on unlabeled sequences from multiple sources, the same frozen encoder supports action classification, dense point generation, joint pose estimation, and future motion prediction on datasets not used for pretraining. Across these tasks, PT cubed outperforms the compared pretraining baselines and achieves competitive or superior performance to task-specific models.

Related papers