Tag
OpenAI hosted its first Robotics Symposium on April 27, 2019, bringing together robotics and machine learning experts to discuss learning robots and demonstrate their humanoid robot hand solving manipulation tasks using vision and reinforcement learning.
OpenAI proposes an asymmetric actor-critic method for robot learning that leverages full state observability in simulators to train policies that operate on partial observations (RGBD images), enabling effective sim-to-real transfer without real-world training data.
A Google DeepMind experiment showed Gemini 4 successfully controlling a humanoid robot to play cornhole by watching videos of humans, practicing through trial and error in the Mojoku simulator, and iteratively fine-tuning based on feedback about where the bags landed. In the end, it sank all seven bags in a row into the hole.