Awesome

AR4 ROS Driver

ROS 2 driver of the AR4 robot arm from Annin Robotics. Tested with ROS 2 Jazzy on Ubuntu 24.04. Also supports ROS 2 Iron using this branch

Supports:

• AR4 (Original version)
• AR4 MK2
• AR4 MK3
• AR4 servo gripper

Features:

• Moveit control (GUI and Python interface example)
• Gazebo simulation
• Hand-Eye calibration

This is a refresh of ar3_core.

Video Demo:

Motion Planning using RViz and Moveit:

Startup, Calibration, and Servo Gripper Demo

Overview

• annin_ar4_description
  • Hardware description of arm & servo gripper urdf.
• annin_ar4_driver
  • ROS interfaces for the arm and servo gripper drivers, built on the ros2_control framework.
  • Manages joint offsets, limits and conversion between joint and actuator messages.
  • Handles communication with the microcontrollers.
• annin_ar4_firmware
  • Firmware for the Teensy and Arduino Nano microcontrollers.
• annin_ar4_moveit_config
  • MoveIt module for motion planning.
  • Controlling the arm and servo gripper through Rviz.
• annin_ar4_gazebo
  • Simulation on Gazebo.

Installation

• Install ROS 2 Jazzy for Ubuntu 24.04
• Clone this repository:

  git clone https://github.com/ycheng517/ar4_ros_driver
  

• Install workspace dependencies:

  rosdep install --from-paths . --ignore-src -r -y
  

• Build the workspace:

  colcon build
  

• Source the workspace:

  source install/setup.bash
  

• Enable serial port access if you haven't already done so:

  sudo addgroup $USER dialout
  

  You will need to log out and back in for changes to take effect.

Firmware Flashing

The Teensy and Arduino Nano sketches provided in annin_ar4_firmware are compatible with the default hardware. To flash it, follow the same procedure as specified in AR4 Robot Setup.

[Optional] Running in Docker Container

A docker container and run script has been provided that can be used to run the robot and any GUI programs. It requires rocker to be installed. Then you can start the docker container with:

docker build -t ar4_ros_driver .

# Adjust the volume mounting and devices based on your project and hardware
rocker --ssh --x11 \
  --devices /dev/ttyUSB0 /ttyACM0 \
  --volume $(pwd):/ar4_ws/src/ar4_ros_driver -- \
  ar4_ros_driver bash

Usage

There are two modules that you will always need to run:

1. Arm module - this can be for either a real-world or simulated arm

   • For controlling the real-world arm, you will need to run the annin_ar4_driver module
   • For the simulated arm, you will need to run the annin_ar4_gazebo module
   • Either of the modules will load the necessary hardware descriptions for MoveIt

2. MoveIt module - the annin_ar4_moveit_config module provides the MoveIt interface and RViz GUI.

The various use cases of the modules and instructions to run them are described below:

MoveIt Demo in RViz

If you are unfamiliar with MoveIt, it is recommended to start with this to explore planning with MoveIt in RViz. This contains neither a real-world nor a simulated arm but just a model loaded within RViz for visualisation.

The robot description, moveit interface and RViz will all be loaded in the single demo launch file

ros2 launch annin_ar4_moveit_config demo.launch.py

Control real-world arm with MoveIt in RViz

Start the annin_ar4_driver module, which will load configs and the robot description:

ros2 launch annin_ar4_driver driver.launch.py calibrate:=True

Available Launch Arguments:

• ar_model: The model of the AR4. Options are mk1, mk2, or mk3. Defaults to mk3.
• calibrate: Whether to calibrate the robot arm (determine the absolute position of each joint).
• include_gripper: Whether to include the servo gripper. Defaults to: include_gripper:=True.
• serial_port: Serial port of the Teensy board. Defaults to: serial_port:=/dev/ttyACM0.
• arduino_serial_port: Serial port of the Arduino Nano board. Defaults to arduino_serial_port:=/dev/ttyUSB0.

⚠️📏 Note: Calibration is required after flashing firmware to the Teensy board, and power cycling the robot and/or the Teensy board. It can be skipped in subsequent runs with calibrate:=False.

Start MoveIt and RViz:

ros2 launch annin_ar4_moveit_config moveit.launch.py

Available Launch Arguments:

• ar_model: The model of the AR4. Options are mk1, mk2, or mk3. Defaults to mk3.
• include_gripper: Whether to include the servo gripper. Defaults to: include_gripper:=True.
• use_sim_time: Make Moveit use simulation time. Should only be enabled when running with Gazebo. Defaults to: use_sim_time:=False.

You can now plan in RViz and control the real-world arm. Joint commands and joint states will be updated through the hardware interface.

NOTE: At any point you may interrupt the robot movement by pressing the E-Stop button on the robot. This would abruptly stop the robot motion! To reset the E-Stop state of the robot use the following command

ros2 run annin_ar4_driver reset_estop.sh <AR_MODEL>

where <AR_MODEL> is the model of the AR4, one of mk1, mk2, or mk3

Control simulated arm in Gazebo with MoveIt in RViz

Start the annin_ar4_gazebo module, which will start the Gazebo simulator and load the robot description.

ros2 launch annin_ar4_gazebo gazebo.launch.py

Start Moveit and RViz:

ros2 launch annin_ar4_moveit_config moveit.launch.py use_sim_time:=true include_gripper:=True

You can now plan in RViz and control the simulated arm.

Hand-Eye Calibration

See ar4_hand_eye_calibration

Tuning and Tweaks

Tuning Joint Offsets

If for some reason your robot's joint positions appear misaligned after moving to the home position, you can adjust the joint offsets in the joint_offsets/ config folder. Select and modify the YAML file corresponding to your AR model to fine-tune the joint positions.

Switching to Position Control

By default this repo uses velocity-based joint trajectory control. It allows the arm to move a lot faster and the arm movement is also a lot smoother. If for any reason you'd like to use the simpler classic position-only control mode, you can set velocity_control_enabled: false in driver.yaml. Note that you'll need to reduce velocity and acceleration scaling in order for larger motions to succeed.

Demo Project (and Python Interface Usage)

See: Tabletop Handybot for a demo of using PyMoveit2 to interface with this driver