The article introduces FootstepNet, an efficient method for bipedal footstep planning and forecasting. It addresses the challenges of designing humanoid locomotion controllers and presents a heuristic-free approach based on Deep Reinforcement Learning. The method is validated through simulation results and real-world deployment on a kid-size humanoid robot during the RoboCup 2023 competition. The content covers the introduction, problem statement, background on RL and DRL, method description, experiments conducted, deployment in RoboCup, conclusions, and references.
Para outro idioma
do conteúdo fonte
arxiv.org
Principais Insights Extraídos De
by Clém... às arxiv.org 03-20-2024
https://arxiv.org/pdf/2403.12589.pdfPerguntas Mais Profundas