NVIDIA Isaac Manipulator
Isaac Manipulator is a collection of foundation models and modular GPU-accelerated libraries that help build scalable and repeatable workflows for dynamic manipulation tasks by accelerating AI model training and task (re)programming. It’s revolutionizing how robotics software developers can leverage customized software components for specific tasks such as machine tending, assembly tasks, etc., enabling manipulation arms to seamlessly perceive and interact with their surroundings.
Sign up below to get notified about the developer preview.
Generic manipulation workflow
Isaac Manipulator simplified end-to-end workflow.
Industry Partners
Adding Next-Level Robot Intelligence
Isaac Manipulator brings new levels of dexterity and modular AI capabilities to robotic arms that face limitations in handling intricate tasks and dynamic environments due to their limited adaptability and manual re-tasking processes for every new scenario.
Highly Accurate & Performant Modules
Access a collection of advanced, modular packages for frictionless kinematics, AI perception, manipulation for industrial robotic arms, and creation of end-to-end workflows on the fly.
Accelerated Robot Motion Generation
NVIDIA cuMotion with a MoveIt extension lets you implement custom algorithmic modules without a large pipeline overhead. cuMotion is built with the help of NVIDIA cuRobo.
Faster Development Time
Speed up robotic task implementations with pretrained foundational models that can estimate and track poses of objects, predict ideal grasp points, run robotic arm trajectory optimizations, and more.
Industry Ready Foundation Models for Manipulators
FoundationPose
This state-of-the-art foundational model for 6D pose estimation and tracking of novel objects was developed by NVIDIA to help you build applications that repeatedly encounter prior unseen objects. NVIDIA’s FoundationPose offers a new method for estimating and tracking the pose of novel objects, working both in situations with or without 3D model existence. It’s instantly applicable to novel objects without any fine-tuning and is robust to diverse objects with challenging properties (textureless, glossy, tiny), as well as in challenging scenes with fast motion or severe occlusions.
NVIDIA cuMotion
cuMotion uses the parallel processing power of NVIDIA GPUs to solve robot motion planning problems at scale by running multiple trajectory optimizations simultaneously to return the best solution.
SyntheticaDETR
This model for object detection in indoor environments allows for faster detection, rendering, and training of new objects. It can also be used as a front-end to pose estimators like FoundationPose, so it can localize objects using 2D bounding boxes before pose estimation.
FoundationGrasp
FoundationGrasp is a transformer model that makes dense grasp predictions given an unknown 3D object asset. The identified grasps can then be executed on the robot with a motion generator such as cuMotion. The model is trained entirely with large-scale synthetic data and currently supports suction and parallel-jaw grasps.
FAQ
The developer preview is planned for Q2, 2024.
Isaac Manipulator for commercial use is not open source. Most components will be released under the Isaac ROS 2 license. For research, the original technologies may be released under open-source or research-limited license, such as cuRobo and FoundationPose.
Yes, cuRobo is open source. Reach out to the NVIDIA research licensing team to request a license if you are planning to go for production with cuRobo.
Latest Product News
From Prototypes to Real-World Systems
NVIDIA’s robotics research team has been at the forefront of bringing research to life for large scale manipulation systems.
Building High-Performance Robotic Applications
NVIDIA Riva
NVIDIA Isaac ROS offers a variety of modular hardware accelerated packages for developers who are seeking AI perception capabilities for building any robot of their choice.
Isaac Perceptor for Autonomous Mobile Robots (AMRs)
Learn more about NVIDIA’s collection of subgraphs, models, and hardware-accelerated packages for AMR AI-based 3D perception.
Get notified about the Isaac Manipulator developer preview.