Amin Abyaneh

PhD candidate in AI and Robotics at McGill and Mila. I want robots that can learn, reason, and act in the messy complexity of the real world. My research spans policy learning, through imitation, reinforcement learning, and diffusion-based methods, and world modeling, building internal representations of how environments evolve over time using autoregressive and diffusion models. I care about grounding these in the mathematics of dynamical systems to bring structure and reliability to learned behavior. Lately I have also been drawn to multimodal learning as a way to give robots richer, more complete pictures of their surroundings.

Papers Blog CV Misc

Latest Updates

Paper Mar 2026

TacFiLM is out — tactile modality fusion for vision-language-action models, in collaboration with NVIDIA Research.

Paper Mar 2026

VOCALoco accepted to IEEE Robotics and Automation Letters (RA-L).

Lecture Mar 2026

Guest lecture at COMP 765 (Robot Learning) on diffusion policies and contraction theory.

Lecture Feb 2026

Guest lecture at COMP 417 (Introduction to Robotics) on imitation learning and dynamical systems.

Paper Jan 2026

Contractive Diffusion Policies (CDP) accepted to ICLR 2026 — heading to Rio de Janeiro, Brazil.

Paper Jan 2025

Contractive Dynamical Imitation Policies accepted to ICLR 2025.

Publications

TacFiLM overview comparing VLA baselines

Tactile Modality Fusion for Vision-Language-Action Models

Charlotte Morissette, Amin Abyaneh, Wei-Di Chang, Anas Houssaini, David Meger, Hsiu-Chin Lin, Jonathan Tremblay, Gregory Dudek

Concept: contraction in diffusion sampling

Contractive Diffusion Policies: Robust Action Diffusion via Contractive Score-Based Sampling with Differential Equations

Amin Abyaneh, Charlotte Morissette, Mohamad Danesh, Anas Houssaini, David Meger, Gregory Dudek, Hsiu-Chin Lin

VOCALoco design overview

VOCALoco: Viability-Optimized Cost-Aware Adaptive Locomotion

Stanley Wu, Mohamad H. Danesh, Simon Li, Hanna Yurchyk, Amin Abyaneh, Anas El Houssaini, David Meger, Hsiu-Chin Lin

SCDS design overview

Contractive Dynamical Imitation Policies for Efficient Out-of-Sample Recovery

Amin Abyaneh*, Mahrokh Boroujeni*, Hsiu-Chin Lin, Giancarlo Ferrari-Trecate

SNDS method overview

Globally Stable Neural Imitation Policies

Amin Abyaneh, Mariana Sosa Guzmán, Hsiu-Chin Lin

PLYDS design overview

Learning Lyapunov-Stable Polynomial Dynamical Systems through Imitation

Amin Abyaneh, Hsiu-Chin Lin

FedCDI framework design

Federated Causal Discovery From Interventions

Amin Abyaneh, Nino Scherrer, Patrick Schwab, Stefan Bauer, Bernhard Schölkopf, Arash Mehrjou