Make a Robot Arm Feel Weightless with NVIDIA Holoscan 5
Web page · https://developer.nvidia.com/blog/make-a-robot-arm-feel-weightless-with-nvidia-holoscan-5
Picture guiding a robot arm by hand while its motors supply just enough torque to support its weight and keep the robot upright. This simple interaction, a...
First seen by Oriel on .
AI summary
This tutorial builds a 400 Hz gravity-compensation loop for reBot Arm using three Holoscan 5 operators, demonstrating inspectable low-jitter robot control and substantially lower tail latency with CPU pinning.
- JointStateSourceOp reads seven motors every 2.5 ms, PinocchioGravityOp computes gravity feedforward with Pinocchio RNEA, and MotorCommandSinkOp validates and sends commands.
- A single-thread partition avoids handoffs, while depth-one DROP_OLDEST queues prevent stale states and commands from accumulating.
- CPU pinning reduced p99.99 latency from 86.6 to 17.2 µs for C++, and from 1,373.5 to 142.6 µs for Python.
- A 100 ms watchdog disables motors and stops the graph if command batches cease; shutdown callbacks also disable the arm.
- The walkthrough supports Jetson Orin Nano, Jetson Thor, and x86 Ubuntu 24.04, with Python and C++ implementations profiled through built-in NVTX annotations in Nsight Systems.
#holoscan #robot control #pinocchio #nsight systems
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| NVIDIA Developer Blog | https://developer.nvidia.com/blog/?p=124655 | https://developer.nvidia.com/blog/make-a-robot-arm-feel-weightless-with-nvidia-holoscan-5/ Read discussion |