NOSTRA: Enabling Robust Robot Imitation via Multimodal Latent Imagination
RA-L, 2026
Vaibhav Saxena
I am a ML Ph.D. student at Georgia Tech, where I work at the Robot Learning and Reasoning Lab (RL2) directed by Prof. Danfei Xu. My research involves generative models for memory 🧠, manipulation 🦾, and vision 👁️. My primary focus is on learning generalizable robot policies using minimal expert supervision.
See more about my professional experience here and also on my CV (Aug, 2025).
News
- [Sept 2026] MetaGenSim and MESA accepted at CoRL 2026 🎉
- [July 2026] NOSTRA accepted at RA-L 🎉
- [April 2026] KinDER accepted at RSS 2026 🎉
- [July 2025] We release the full MimicLabs framework, consisting of data generation code, simulation environments, and datasets.
- [Jan 2025] MimicLabs accepted at ICLR 2025 🎉
- [Aug 2024] Our paper on “C3DM: Constrained-Context Conditional Diffusion Models for Imitation Learning” got accepted at TMLR 2024 🎉
- [Mar 2024] We will be organizing the “Data Generation for Robotics Workshop” at RSS 2024 in Delft, Netherlands!
- [Feb 2023] I will be returning to Autodesk Research as a Research Intern during Summer 2023
- [Jan 2023] Our paper on “Generalizable Pose Estimation using Implicit Scene Representations” got accepted at ICRA 2023 🎉
- [Dec 2022] I will be presenting our paper on “Guiding Exploration Towards Impactful Actions” at the Deep RL Workshop in NeurIPS 2022
- [Feb 2022] I will be spending the Summer of 2022 as a Research Intern at Autodesk Research
- [Dec 2021] I will be presenting our paper on “Clockwork Variational Autoencoders” at NeurIPS 2021 [tweet]
- [Aug 2021] Joining the ML Ph.D. program at Georgia Tech!
- [Apr 2021] Clockwork VAEs featured on Towards Data Science!
Research (representative works highlighted)
KinDER: A Physical Reasoning Benchmark for Robot Learning and Planning
RSS, 2026
What Matters in Learning from Large-Scale Datasets for Robot Manipulation
ICLR 2025
MimicTouch: Leveraging Multi-modal Human Tactile Demonstrations for Contact-rich Manipulation
CoRL 2024
C3DM: Constrained-Context Conditional Diffusion Models for Imitation Learning
Workshop on Deployable Robotics, CoRL 2023
TMLR 2024
Generalizable Pose Estimation using Implicit Scene Representations
ICRA, 2023
Guiding Exploration Towards Impactful Actions
Deep Reinforcement Learning Workshop
NeurIPS, 2022
Clockwork Variational Autoencoders
NeurIPS, 2021
Dyna-AIL: Adversarial Imitation Learning by Planning
Beyond “Tabula Rasa” in Reinforcement Learning Workshop
ICLR, 2020
Service
Reviewer for
- ML Conferences: NeurIPS, ICML, ICLR
- Robotics Conferences: CoRL, RSS, ICRA
- Journals: RA-L, TMLR