Sim2Real Reinforcement Learning for Soccer skills

Jonathan Spraggett

Sim2Real Reinforcement Learning for Soccer skills

Robotics 2025-12-17 v1 Machine Learning

Authors: Jonathan Spraggett

Abstract

This thesis work presents a more efficient and effective approach to training control-related tasks for humanoid robots using Reinforcement Learning (RL). The traditional RL methods are limited in adapting to real-world environments, complexity, and natural motions, but the proposed approach overcomes these limitations by using curriculum training and Adversarial Motion Priors (AMP) technique. The results show that the developed RL policies for kicking, walking, and jumping are more dynamic, and adaptive, and outperformed previous methods. However, the transfer of the learned policy from simulation to the real world was unsuccessful, highlighting the limitations of current RL methods in fully adapting to real-world scenarios.

Keywords

sim-to-real transfer reinforcement learning imitation learning

Cite

@article{arxiv.2512.12437,
  title  = {Sim2Real Reinforcement Learning for Soccer skills},
  author = {Jonathan Spraggett},
  journal= {arXiv preprint arXiv:2512.12437},
  year   = {2025}
}

Comments

Undergrad Thesis

Related papers

View all related →