Sundaralingam, who built the first 100 ms cuRobo prototype in 2022, is releasing NVIDIA's 4.4 ms planner with Adithyavairavan Murali and Stan Birchfield.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [Aligned News — AI Intelligence](https://x.com/balakumar_/status/2118408305857228876)
Why it matters #
Motion planning sits between a robot seeing an object and moving to grasp it. Cutting that step to 4.4 ms on stated hardware, while respecting payload and torque limits, reduces the time spent planning a path. Releasing it under Apache 2.0 through NVlabs gives research teams and robotics companies a planner they can inspect, benchmark on their own arms, and build products around.
Balakumar Sundaralingam (@balakumar_) started cuRobo with a planner that took 100 ms. "In 2022, our first cuRobo prototype took 100 ms and consumed the whole GPU," he wrote in a thread on X on November 5th. Four years of kernel work later, he is open-sourcing cuRoboV2-Fast under Apache 2.0, with collision-free, torque-limited motion planning in 4.4 ms.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.