Lightwheel has launched RoboFinals RL Environments, a reinforcement-learning simulation environment that provides continuously interactive and physically credible feedback for training large models, with the RoboFinals Benchmark offering large-scale evaluation. The two share task semantics, scene assets and capability definitions so that training and evaluation target the same operational requirements, while training tasks, reward-tuning processes and independent evaluation instances remain separate to preserve an independent basis for testing learning outcomes.
Lightwheel launches RoboFinals RL Environments for robot model training
Related Articles
Lightwheel, Unitree sign strategic partnership on physical AI, robot deployment
Lightwheel and Peritas AI Announce Partnership for $56M Multi-Phase Project to Bring Robots into Perioperative Workflows
AGIBOT Lingxi X2 refreshed edition offers up to 47 degrees of freedom
ACE ROBOTICS unveils ACE-CaP-0 for reusable robot skills
Shanghai construction robot maker WeIBuild raises over 100 mln yuan in Series B
OpenAI disbands standalone robotics team after Astra breakthrough
Boston Dynamics shows Atlas humanoid robot fitted with wheels
D-Robotics, Infinigence AI partner on open-source embodied AI infrastructure


AGIBOT
Figure AI
UBTECH