Kimodo
A scalable and controllable kinematic diffusion model for humanoid motion generation.
Overview
Our mission is to develop AI-driven methods for synthesizing high-quality 3D human motion. We aim to create motion generation tools to facilitate any downstream applications that may involve human motion, including robotics, animation, and synthetic data generation.
Key Research Areas
Human Motion Generation • Physics-Based Control • Human Motion Understanding and Perception
Featured Research Projects
ProtoMotions
A framework for training physically simulated humanoids to reproduce the full range of human motor skills.
Human Motions from Video
Exploring ways to leverage in-the-wild and generated videos to collect motion data at scale.
Resources & Publications
- Kimodo: Scaling Controllable Human Motion Generation - Tech Report 2026
- Generative AI Body Motion - GDC 2025
- GENMO: A Generalist Model for Human Motion - ICCV 2025
- MaskedMimic: Unified Physics-Based Character Control Through Masked Motion Inpainting - SIGGRAPH Asia 2024
- TeSMO: Generating Human Interaction Motions in Scenes with Text Control - ECCV 2024
- SuperPADL: Scaling Language-Directed Physics-Based Control with Progressive Supervised Distillation - SIGGRAPH 2024
- STMC: Multi-Track Timeline Control for Text-Driven 3D Human Motion Generation - CVPR 2024
- Learning Physically Simulated Tennis Skills from Broadcast Videos - SIGGRAPH 2023
- Synthesizing Physical Character-Scene Interactions - SIGGRAPH 2023
- CALM: Conditional Adversarial Latent Models for Directable Virtual Characters - SIGGRAPH 2023
- Trace and Pace: Controllable Pedestrian Animation via Guided Trajectory Diffusion - CVPR 2023
- PADL: Language-Directed Physics-Based Character Control - SIGGRAPH Asia 2022
- ASE: Large-Scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters - SIGGRAPH 2022