China’s PrimeBOT has launched PrimeSim, a world-model simulator for robot training, and said it topped the WorldSimProbe LIBERO leaderboard, setting a new benchmark for robot simulation performance. PrimeSim builds an interactive virtual world in which robots can trial-and-error and learn continuously, enabling efficient policy validation and reinforcement learning. In tests, reinforcement learning using PrimeSim raised policy success rates to 73.2% from 61.7%.
China's PrimeBOT unveils PrimeSim world model, tops WorldSimProbe LIBERO leaderboard
Related Articles
PrimeBOT to debut Q1, T1 robots at Shanghai multi-site launch event >
PrimeBOT launches personal robots at IFA Berlin, targets home market with affordable modular designs
D-Robotics, Unitree team up to launch embodied AI education solution
China's Zhicheng AI raises several hundred million yuan in funding round
Apple, Alibaba partner to bring Qwen to Apple Intelligence in China
OpenAI cuts fundraising target to about $20 bln as tech selloff deepens AI spending worries
Shanghai targets 8,000 humanoid robots in factories by 2028
GENISOM AI, Tencent Cloud sign strategic cooperation pact on embodied intelligence

AGIBOT
Figure AI
UBTECH