cuRobo: CUDA Accelerated Robot Library

[API] [GTC Talk] [Research] [Code] [Arxiv] [PDF]

This is a Preview Release highlighting the results obtained in our Technical Report

cuRobo is shipping as NVIDIA cuMotion with NVIDIA Isaac Manipulator in Q2 2024!

For business inquiries, please submit this form: NVIDIA Research Licensing

Updates

• April 2024 (v0.7.2): Added re-timing to change planned trajectory speed between planning calls. Higher-level API documentation. Improvements for smoother trajectories.

• April 2024 (v0.7.1): Added mimic joint support, Improved joint space planning, added API documentation for IKSolver. Also added high precision (<1mm) option to IKSolver and MotionGen.

• March 2024 (v0.7.0): ESDF voxel grid collision checker, partial support for Isaac Sim 2023.1.1, improved collision kernels.

• Feb 2024 (v0.6.3): New: Constrained Planning, Robot Segmentation, Update Locked JointState.

• Dec 2023 (v0.6.2): New features and quality improvements.

• Nov 2023 (v0.6.1): Added support for Isaac Sim 2023.1.0, x86, aarch64, and isaac sim dockerfiles.

• Oct 2023 (v0.6.0): Released cuRobo Technical Report and Source Code.

See CHANGELOG for details.

Overview

cuRobo is a CUDA accelerated library containing a suite of robotics algorithms that run significantly faster than existing implementations leveraging parallel compute. cuRobo currently provides the following algorithms: (1) forward and inverse kinematics, (2) collision checking between robot and world, with the world represented as Cuboids, Meshes, and Depth images, (3) numerical optimization with gradient descent, L-BFGS, and MPPI, (4) geometric planning, (5) trajectory optimization, (6) motion generation that combines inverse kinematics, geometric planning, and trajectory optimization to generate global motions within 30ms.

cuRobo generates motions for a UR10 within 100ms on a NVIDIA Jetson Orin.

cuRobo performs trajectory optimization across many seeds in parallel to find a solution. cuRobo’s trajectory optimization penalizes jerk and accelerations, encouraging smoother and shorter trajectories.

cuRobo leverages nvblox for collision avoidance with obstacles from a Depth camera.

Comparison of cuRobo’s motion generation on the left to a BiRRT planner for the motion planning phases in a pick and place task.

cuRobo can constrain different axes and allow for planning in the remaining axes leveraging trajectory optimization.

Example motions generated by cuRobo on the motionbenchmaker and motion policy networks datasets.

• Research
  • Technical Report
    • cuRobo: Parallelized Collision-Free Minimum-Jerk Robot Motion Generation
    • Citation
  • Research using cuRobo

• Get Started
  • Installation
    • System Requirements
    • Library Installation
    • Install for use in Isaac Sim
    • Install nvblox (optional)
      • Installing nvblox for CXX11_ABI and pytorch dockers
      • Installing nvblox for PRECXX11_ABI and Isaac Sim
    • Installing on NVIDIA Jetson (aarch64)
  • Basics
  • Using in Python
    • Forward Kinematics
    • Collision Checking
    • Inverse Kinematics
    • Motion Generation
    • Model Predictive Control (MPC)
  • Using with Isaac Sim
    • Usd Interface
    • Isaac Sim Basics
    • Collision Checking
    • Motion Generation
    • Reachability using Batched Collision-Free Inverse Kinematics
    • Model Predictive Control with MPPI
    • Multi-Arm Reacher
  • Collision World Representation
    • World Representation
    • World Collision Checkers
      • Batched Environments
    • Collision Checking Implementation
      • Collision Metric
    • Geometry Approximation to Spheres
  • Using with Depth Camera
    • Demos with an existing map
      • Collision Checking
      • Motion Generation
      • Model Predictive Control
    • Demos with online mapping
      • Collision Checking
      • Motion Generation
      • Model Predictive Control
  • Using in a Neural Network
  • Benchmarks & Profiling
    • Benchmarks
      • Latest Motion Generation Results
      • Latest Inverse Kinematics Results
      • Running Benchmarks
      • Kinematics & Collision Checking
      • Inverse Kinematics
      • Motion Generation & Geometric Planning
    • Profiling
  • Docker Development
    • Building your own docker image with cuRobo
    • Build Warp for NVIDIA Jetson (Deprecated)
  • Known Issues
    • cuRobo
    • Python
    • nvblox
    • Isaac Sim
  • Python API

• Advanced Examples
  • Batched Environments
    • Collision Checking
    • Motion Generation
  • Block Stacking
    • Attaching Object to Robot
  • Constrained Planning
    • Isaac Sim Demo
    • Grasp Approach Vector
  • Robot Segmentation

• Tutorials
  • Configuring a New Robot
    • Prequisite
    • Robot Configuration File
    • Tutorial with a UR5e robot
      • Prepare assets
      • Create Configuration File
      • Robot Collision Representation
      • Self Collision Configuration
      • Additional Configurations
      • Test Robot Configuration

• Developer Notes
  • Supported Robots
  • Numerical Optimization
  • Performance Tuning
    • Improving Start Time
  • CUDA Notes
    • Developing CUDA kernels
    • Tensor IO Buffers
    • Batched Environment Collision Checking
  • USD for Robot and World Representation
    • Reading World from USD
  • Environment Variables