In the research area of robotics, there exists a huge gap between academia and industry. There are mainly two bottlenecks: (1) it costs too much to conduct large-scale robotic system tests for academia to examine the generalization capability of robotic algorithms, and (2) the evaluation criteria of robotics is not aligned between academia and industry due to the high dependency on robot hardware configuration. Therefore, we hope to bridge this gap and the first step we have taken is hosting the RoboMaster University Sim2Real Challenge at ICRA2022. Our group aims to build a cloud-based robotic sim2real platform where we may solve such problems by investigating simulation-to-reality gaps and performing explainable domain transfer learning which is supposed to shorten deployment duration in robotics. We hope to attract researchers around the world to make use of our platform and contribute to Sim2Real-based robotics, after which the research progress in robotics could hopefully be accelerated.
AIR, Tsinghua University, in collaboration with the RoboMaster Organizing Committee, will launch the 2nd RoboMaster University Sim2Real Challenge (RMUS) at ICRA 2023. In this challenge, the organizer will provide a downloadable simulator and a cloud-based robotic game field. Throughout the process of completing a specific rearrangement task, participants can develop navigation and grasping algorithms with the simulator and can remotely access the physical robot in the game field for a limited number of verifications. Over the past five years, RoboMaster challenges have been held uninterruptedly at ICRA and have attracted more than 100 teams worldwide from universities and robotics labs. Last year, 117 teams from MIT, CMU, ETH, ICL, TUM, NUS, Tsinghua, and other universities participated in the first RMUS. For ICRA 2023, we propose our new RMUS with an additional adversarial mechanism and look forward to further expanding our impact.