Papers

These are my most recent papers as first author. For the full list, check my Google Scholar.

CopyQuad: Quadrupedal Skill Transfer from a Single Side-View Video Without a Depth Model

Preprint

CopyQuad skips the depth-estimation stage that video-to-robot pipelines usually need. For side-view footage of a walking or trotting animal, it pins the one axis the camera can’t see to the robot’s own stance width, then measures what that costs against real depth models and motion capture.

Qualitative comparison of PPO, FastTD3, SAC, and SMM-AC on the Franka Lift and ANYmal-D Rough sim-to-real tasks

Regularised RL Improves Sim-to-Real Transfer

Preprint

Regularised RL has long promised robustness guarantees, but weakly regularised algorithms like PPO still dominate sim-to-real pipelines. We derive SMM-AC, a continuous-control actor-critic built on the soft mellowmax value function, and show on a Franka arm and an ANYmal-D quadruped that regularisation translates into real sim-to-real gains.

Morphology-Conditioned World Model for Cross-Embodiment Quadrupedal Locomotion

CoRL 2026

QWM conditions one world model on scale-invariant physical features and trains every policy inside it. Given the same morphology spec, a model-free policy collapses on unseen robots while QWM transfers zero-shot to new quadrupeds, in simulation and on hardware.

Overview of DriftQL: sampling states and noise, forming positives and negatives, computing the conditional drift field, and the drift plus Q-learning loss

Drift Q-Learning

NeurIPS 2026

DriftQL learns a single drift field, balancing attraction toward the dataset with repulsion for diversity, so an offline policy can act in one forward pass, no denoising chain or ODE solver required. It matches diffusion and flow policies on D4RL and OGBench and stays far more robust when the data is corrupted.