Research publications by Agon Serifi, Disney Research Robotics https://aserifi.github.io/ 2026-04-06T00:00:00Z Agon Serifi agon.serifi@disneyresearch.com https://arxiv.org/abs/2603.16536 2025-03-01T00:00:00Z

This paper presents Kamino, a GPU-based physics solver for massively parallel simulation of heterogeneous, highly-coupled mechanical systems. Built in Python with NVIDIA Warp and integrated into the Newton framework, Kamino natively supports complex topologies with kinematic loops, avoiding tree-based approximations.

https://arxiv.org/abs/2512.16705 2025-01-01T00:00:00Z

In this work, we bring Olaf, an animated character, to life in the physical world, relying on reinforcement learning guided by animation references for control.

https://arxiv.org/abs/2511.10635 2025-01-01T00:00:00Z

This paper presents a method for learning controlled, soft falls for bipedal robots. Rather than preventing falls, we focus on minimizing physical damage and achieving desired end poses through a robot-agnostic reward function that balances impact minimization with pose control during reinforcement learning.

https://doi.org/10.1145/3721238.3730656 2025-01-01T00:00:00Z

This paper presents AMOR, a method for adaptive character control through multi-objective reinforcement learning. A single policy is trained conditioned on reward weights and motion context, enabling post-training tuning of reward trade-offs and zero-shot adaptation to diverse tracking behaviors.

https://arxiv.org/abs/2504.02724 2025-01-01T00:00:00Z

This paper presents a method for autonomously controlling expressive human-robot interactions by training a diffusion-based imitation model on real operator data. The resulting system predicts both continuous and discrete control inputs while conditioning on human pose.

https://doi.org/10.1145/3680528.3687626 2024-12-01T00:00:00Z

This paper presents a method for aligning text-conditioned kinematic motion generators with the capabilities of robotic characters. Our approach combines motion generation with physics-based character control using a critic that acts as a reward surrogate.

https://doi.org/10.1007/978-3-031-73016-0_1 2024-10-01T00:00:00Z

A new class of Transformer models that eliminate the need for positional encoding. Spline-based Transformers embed an input sequence of elements as a smooth trajectory in latent space, significantly outperforming traditional transformer models in efficiency and accuracy.

https://doi.org/10.1111/cgf.15175 2024-07-01T00:00:00Z

This paper introduces a two-stage framework for transferring expressive and dynamic motions onto robotic characters. Starting from a large corpus of human motion capture data, we first learn a latent space representation, then use it as a prior for training a tracking controller through reinforcement learning.

https://doi.org/10.1109/LRA.2023.3271812 2023-10-01T00:00:00Z

This paper introduces a modular approach for neural augmentation of simulation states, improving robot simulation accuracy by targeting individual building blocks such as actuators and rigid bodies separately.