Accelerated Robot Training through Simulation with ROS and Gazebo Risto Kojcev, Alejandro Hernández Cordero, Asier Bilbao Calvo, Irati Zamalloa Ugarte, Yue Leire Erro Nuin, Víctor Mayoral Vilches AI Team @ Erle Robotics We investigate the state of the art deep-learning methods, and apply them to modular robots. SSSooouuurrrccceee::: EEErrrllleee RRRooobbboootttiiicccsss AI Infrastructure : Algorithms powered by HRIM powered by AI Two unlinked worlds Sadeghi, Fereshteh, et al. CVPR 2018 Two unlinked worlds o b e z a g _ m Sadeghi, Fereshteh, et al. CVPR 2018 y g GYM_GAZEBO : a Link between AI and Robotic Methods environment API Robot toolbox RL Algorithm and libraries Simulator Github stats GYM_GAZEBO is Open Source. Try it! Watching Stars https://github.com/erlerobot/gym-gazebo 42 395 GYM_GAZEBO : Community Projects   Nicola Ristè Institut de Robòtica i Informàtica https://youtu.be/h84IkUCze44 An “end to end” solution for modular robots Reinforcement Learning requires long training time… …this is me waiting for results !