MoveIt Inverse Kinematics Benchmarking

Introduction

Inverse Kinematics (IK) serves as a foundational element in robotic systems, facilitating purposeful interactions with the surrounding environment. It empowers robots to achieve specific poses and reach target locations with precision. Despite the importance of IK solvers in robotic planning and control, choosing the right one can be a complex decision. Different IK solvers offer unique strengths and weaknesses, raising the need to conduct a performance evaluation for specific use-cases.

This ik_benchmarking package utilizes ROS 2 and MoveIt 2 to offer a suite of benchmarking utilities designed to aid the evaluation of IK solvers. This tutorial is crafted to walk you through the installing the package, configuring IK solvers for benchmarking, running the necessary scripts for data collection and visualization of the results for easier analysis.

In addition, the architectural components of the package are outlined with key classes that enable its functionality. Towards the end, we discuss potential future improvements, ensuring that the package remains aligned with emerging needs and technologies.

Installation

In the following steps, the ik_benchmarking is assumed to be installed in the ws_moveit2 workspace as it is closely connect with MoveIt 2, but feel free to use your own workspace. The default example of ik_benchmarking uses KDL, bio_ik, and TRAC_IK solvers.

1. Clone the repository and install dependencies

    cd ~/ws_moveit2/src
    git clone https://github.com/PickNikRobotics/ik_benchmarking.git
    vcs import < ik_benchmarking/.repos

    # Install package dependencies
    rosdep update
    rosdep install -r --from-paths . --ignore-src --rosdistro $ROS_DISTRO -y
    

1. Build the workspace as follows

    cd ~/ws_moveit2
    colcon build --symlink-install
    

Note: Including the --symlink-install flag is advantageous as it allows you to make changes to the package files without requiring a complete rebuild of the workspace. This applies only to files that are interpreted at run time, like YAML, Python scripts, etc.

1. Source the Workspace

    source install/setup.bash
    

With these steps completed, we are now set to dive into the configuration of the IK solvers for benchmarking purposes.

Usage

Configuration via ik_benchmarking.yaml

Before initiating the benchmarking process, ensure that the ik_benchmarking.yaml configuration file is tailored according to your needs. This file allows users to define various settings such as the MoveIt configuration package to load the robot model, the planning group for the robot pre-defined within the MoveIt configuration package, the sample size, and the IK solvers intended for testing.

Below is a breakdown of the default example configuration file’s structure:

moveit_config_pkg: moveit_resources_panda_moveit_config
robot_name: moveit_resources_panda
planning_group: panda_arm
sample_size: 10000
random_seed: 12345
ik_timeout: 0.1
ik_iteration_display_step: 1000

ik_solvers:
  - name: KDL
    kinematics_file: kdl_kinematics.yaml

  - name: TRAC_IK
    kinematics_file: trac_ik_kinematics.yaml

  - name: bio_ik
    kinematics_file: bio_ik_kinematics.yaml

Key Components

• moveit_config_pkg: Specifies the MoveIt configuration package for the robot arm you are benchmarking. For example, moveit_resources_panda_moveit_config is used for the Panda robot arm. By convention, the MoveIt configuration packages are named robot_moveit_config or moveit_resources_robot_moveit_config. Your robot’s MoveIt config package should follow this convention.

File truncated at 100 lines see the full file