· Product launches· 日本語で読む
Ambi Robotics unveils physical AI model CARGO, surpassing 75% packing density with simulation-based reinforcement learning
Ambi Robotics on Aug. 26, 2026, announced CARGO, a physical AI model for industrial package loading. Trained with Sim2Real reinforcement learning, it achieved packing density above 72.5% and more than 340 sorts per hour in real customer operations. The global package-loading market is estimated at more than $400 billion annually.
What happened
Ambi Robotics on Aug. 26, 2026, announced CARGO, a physical AI model for industrial package loading. According to the company’s blog, CARGO was trained with Sim2Real reinforcement learning and achieved packing density above 72.5% and more than 340 sorts per hour in real customer operations.
Why it matters
In warehouse box-stacking operations, a challenge is deciding where to place the next package while taking into account packages that arrive first. Ambi Robotics says it aims to improve packing density even as box arrangements change by applying CARGO, trained in simulation, to robots in real-world operations.
Sources
Translated from the Japanese original: /news/ambi-robotics-butsuri-ai-moderu-cargo-wo-happyou-tsumi-tsuke-438278. The Japanese version is authoritative if they differ.