dora-moveit2

A reusable robot motion planning library for the dora-rs dataflow framework. Provides MoveIt-style IK solving, collision-aware OMPL planning, trajectory execution, and a high-level MoveGroup API — for any robot arm.

Features

• MoveGroup API — High-level interface similar to ROS MoveIt's MoveGroupCommander
• IK Solver — TracIK-inspired multi-start inverse kinematics
• Motion Planner — RRT-Connect (OMPL) with geometric collision detection
• Planning Scene — Obstacle management and scene state broadcasting
• Trajectory Executor — Quintic spline interpolation with smooth execution
• Robot-agnostic — Bring your own robot config via ROBOT_CONFIG_MODULE env var

Quick Start

# Install the library
pip install -e dora_moveit/

# Install an example app
pip install -e examples/move_group_demo/

# Run the UR5e pick-and-place example
cd examples/move_group_demo
dora up
dora start dataflows/ur5e_example_mujoco.yml

Examples

Example	Robot	Description
move_group_demo	UR5e 6-DOF + Robotiq 2F-85	Pick-and-place demo with MoveGroup API — named poses, joint goals, Cartesian paths, collision objects
hunter_with_arm	Hunter SE + GEN72	Mobile robot with arm for multi-view inspection

See the examples README for details and a guide to creating your own.

Repository Structure

dora_moveit/           # Library package (pip install -e dora_moveit/)
  dora_moveit/
    config.py          # RobotConfig protocol + load_config()
    ik_solver/         # IK solver operators
    motion_planner/    # OMPL planner + planning scene
    trajectory_execution/  # Trajectory executor
    collision_detection/   # Collision checking
    workflow/          # MoveGroup API + motion commander
examples/              # Example applications
  move_group_demo/     # ROS MoveIt-style MoveGroup API example
  hunter_with_arm/     # Hunter SE mobile robot + GEN72 7-DOF arm
dora-mujoco/           # MuJoCo simulation node

Usage

MoveGroup API (ROS MoveIt equivalent)

from dora_moveit.workflow.move_group import MoveGroup

group = MoveGroup("ur5e")
scene = group.get_planning_scene_interface()

# 1. Named pose
group.set_named_target("home")
group.go(wait=True)

# 2. Joint-space goal (6-DOF for UR5e)
group.go([1.57, -0.785, 1.57, -1.57, -1.57, 0.0], wait=True)

# 3. Cartesian pose goal (IK solved internally)
group.set_pose_target([0.35, -0.25, 0.20, 0, 0, 0])
group.go(wait=True)

# 4. Cartesian path (straight line in workspace)
waypoints = [[0.35, -0.25, 0.20, 0, 0, 0], [0.35, 0.25, 0.20, 0, 0, 0]]
trajectory, fraction = group.compute_cartesian_path(waypoints, eef_step=0.01)
group.execute(trajectory, wait=True)

# 5. Collision objects
scene.add_box("obstacle", [0.4, 0.0, 0.3], [0.1, 0.1, 0.3])
group.set_named_target("home")
group.go(wait=True)  # planner avoids the box
scene.remove_world_object("obstacle")

group.shutdown()

Robot Config Injection

Library operators are robot-agnostic. Define a config class for your robot:

# my_app/config/my_robot.py
import numpy as np

class MyRobotConfig:
    NUM_JOINTS = 6  # or 7 for 7-DOF arms
    JOINT_LOWER_LIMITS = np.array([...])
    JOINT_UPPER_LIMITS = np.array([...])
    LINK_TRANSFORMS = [...]  # per-joint axis, xyz, rpy
    COLLISION_GEOMETRY = [...]
    HOME_CONFIG = np.array([...])
    SAFE_CONFIG = np.array([...])
    NAMED_POSES = {"home": np.array([...])}

Set the env var in your dataflow YAML:

- id: planner
  path: ../my_app/nodes/planner.py
  env:
    ROBOT_CONFIG_MODULE: "my_app.config.my_robot"

See dora_moveit/dora_moveit/config.py for the full RobotConfig protocol and examples/ for a complete working example.

Deployment Guide — MuJoCo Demos

Prerequisites

• Python 3.9+
• dora-rs CLI installed (cargo install dora-cli or download a release binary)
• A display server (X11/Wayland) for the MuJoCo viewer

1. UR5e Pick-and-Place Demo

The primary demo — a UR5e 6-DOF arm with Robotiq 2F-85 gripper performing pick-and-place of a red ball. Uses official MuJoCo Menagerie meshes.

# 1. Install packages
pip install -e dora_moveit/
pip install -e dora-mujoco/
pip install -e examples/move_group_demo/

# 2. Launch
cd examples/move_group_demo
dora up
dora start dataflows/ur5e_example_mujoco.yml

# 3. Stop when done
dora stop

Standalone pick-and-place visualization (no dora required):

cd examples/move_group_demo/models
mjpython pick_and_place_demo.py

Dataflow nodes: mujoco_sim → planning_scene → planner → ik_solver → trajectory_executor → user_node

The user_node runs through all 5 MoveGroup demos automatically. The MuJoCo viewer shows the UR5e arm moving in real time.

Troubleshooting:

• If nodes fail with ModuleNotFoundError, ensure all three packages are installed in the same Python environment that dora uses. You can set PYTHONPATH in each node's env: block in the dataflow YAML to point to your site-packages directory.
• If you see version mismatch: message format vX is not compatible with vY, make sure the dora-rs Python package version matches your dora CLI version (dora --version).

2. Hunter with ARM Demo (Mobile Robot + Arm)

Hunter SE mobile platform with a GEN72 arm mounted on top, operating in an industrial pipe inspection scene.

# 1. Install packages
pip install -e dora_moveit/
pip install -e dora-mujoco/
pip install -e examples/hunter_with_arm/

# 2. Launch MoveGroup API demo
cd examples/hunter_with_arm
dora up
dora start dataflows/movegroup_mujoco.yml

# 3. Stop when done
dora stop

Available dataflows:

Dataflow	Description
movegroup_mujoco.yml	MoveGroup API demo — arm moves through named poses, joint goals, and planned trajectories
hunter_arm_mujoco.yml	Choreographed multi-view capture — arm sequences through inspection positions
gen72_mujoco.yml	Standalone GEN72 arm simulation (no Hunter SE vehicle)
gen72_real.yml	Physical GEN72 arm via Realman SDK (requires hardware)

Dataflow nodes (movegroup): mujoco_sim → planning_scene → planner → ik_solver → trajectory_executor → vehicle_controller → user_node

The demo automatically:

1. Moves the arm to home position
2. Turns the arm right
3. Arches over a pipe (capture position)
4. Returns to home
5. Executes a planned trajectory
6. Idles when complete

Troubleshooting:

• Verify the MODEL_NAME path in the dataflow YAML points to the correct .xml file under examples/hunter_with_arm/models/.
• The Hunter SE model (hunter_se_with_arm_and_pipes.xml) includes the vehicle, arm, and industrial pipe scene. If you only want the arm, use gen72_mujoco.yml instead.

Environment Variable Setup (Alternative to pip install)

If you prefer not to use pip editable installs, set PYTHONPATH directly:

export PYTHONPATH=/path/to/dora_moveit:/path/to/dora-mujoco:/path/to/examples/hunter_with_arm:$PYTHONPATH

Or add PYTHONPATH to each node's env: block in the dataflow YAML:

env:
  ROBOT_CONFIG_MODULE: "hunter_arm_demo.config.gen72"
  PYTHONPATH: "/path/to/your/site-packages"

Adding a New Robot

See the examples README for a step-by-step guide to creating a new example application with your own robot.

License

MIT