prometheus DiffDrive Robot

https://github.com/bruk/prometheus/assets/

About

This repository contains a Gazebo and Isaac Sim simulation for a differential drive robot, equipped with an IMU, a depth camera, stereo camera and a 2D LiDAR. The primary contriution of this project is to support multiple ROS and Gazebo distros. Currently, the project supports the following versions -

1. ROS2 Humble/Jazzy with Gazebo Harmonic

Each of the following sections describes depedencies, build and run instructions for each of the above combinations

ROS2 Humble/Jazzy with Gazebo Harmonic

Dependencies

• AWS RoboMaker Small House World ROS package

sudo apt-get install ros-humble-ros-gzharmonic

# From the root directory of the workspace. This will install everything mentioned in package.xml
rosdep install --from-paths src/prometheus --ignore-src -r -y

Build

colcon build --packages-select prometheus

Run

To launch the robot in Gazebo,

ros2 launch prometheus gz.launch.py

To view in rviz,

ros2 launch prometheus rviz.launch.py

Configuration

The launch file accepts multiple launch arguments,

ros2 launch prometheus gz.launch.py \
 arm_enabled:=True \
 camera_enabled:=True \
 stereo_camera_enabled:=False \
 two_d_lidar_enabled:=True \
 position_x:=0.0 \
 position_y:=0.0  \
 orientation_yaw:=0.0 \
 odometry_source:=world \
 world_file:=small_warehouse.sdf

Mapping with SLAM Toolbox

SLAM Toolbox is an open-source package designed to map the environment using laser scans and odometry, generating a map for autonomous navigation.

NOTE: The command to run mapping is common between all versions of gazebo.

To start mapping:

ros2 launch prometheus mapping.launch.py

Use the teleop twist keyboard to control the robot and map the area:

ros2 run teleop_twist_keyboard teleop_twist_keyboard cmd_vel:=/prometheus/cmd_vel

To save the map:

cd src/prometheus/config
ros2 run nav2_map_server map_saver_cli -f map

Using Nav2 with prometheus

Nav2 is an open-source navigation package that enables a robot to navigate through an environment easily. It takes laser scan and odometry data, along with the map of the environment, as inputs.

NOTE: The command to run navigation is common between all versions of gazebo and Isaac sim.

To run Nav2 on prometheus:

ros2 launch prometheus nav2.launch.py

Arm Control Interface

The SO-100 5-DOF arm is controlled via a unified ROS2 interface using Pink (Pinocchio-based IK). The pink_ik_node exposes action servers and services for arm movement, state queries, and gripper control.

Arm Controllers

The arm uses ros2_control with the following controllers (loaded automatically by the launch file):

Controller	Type	Purpose
joint_state_broadcaster	JointStateBroadcaster	Publishes joint states
joint_trajectory_controller	JointTrajectoryController	Position control for 5 arm joints
gripper_controller	GripperActionController	Gripper open/close

Arm joints: Shoulder_Rotation, Shoulder_Pitch, Elbow, Wrist_Pitch, Wrist_Roll

Dependencies

pip install pin-pink quadprog "numpy<2"

Note: numpy<2 is required because the ROS Jazzy system Pinocchio is compiled against NumPy 1.x. The pin-pink package pulls NumPy 2.x by default, which causes a crash.

Running

# Terminal 1: Launch simulation
ros2 launch prometheus gz.launch.py

# Terminal 2: Launch arm control node (loads controllers automatically)
ros2 launch prometheus pink_ik.launch.py

ROS2 Interface

Action: arm/move_to_pose (prometheus/action/MoveArm)

Unified arm movement with 5 modes:

Mode	Description	Required Fields
cartesian	Move EE to absolute Cartesian pose	target_pose
relative	Move EE by offset from current position	offset (dx, dy, dz)
named	Move to a named joint configuration	pose_name (home, ready, tucked)
home	Return to home position	none
joints	Move to direct joint angles	joint_positions (5 floats, radians)

Feedback provides progress (0-1), position_error (meters), and phase (solving/executing).

Example:

# Cartesian goal
ros2 action send_goal /arm/move_to_pose prometheus/action/MoveArm \
  "{mode: 'cartesian', target_pose: {header: {frame_id: 'base_footprint'}, pose: {position: {x: 0.3, y: 0.0, z: 0.3}, orientation: {w: 1.0}}}}"

# Relative move (10cm down)
ros2 action send_goal /arm/move_to_pose prometheus/action/MoveArm \
  "{mode: 'relative', offset: {z: -0.1}}"

# Named pose
ros2 action send_goal /arm/move_to_pose prometheus/action/MoveArm \
  "{mode: 'named', pose_name: 'home'}"

# Home
ros2 action send_goal /arm/move_to_pose prometheus/action/MoveArm "{mode: 'home'}"

Gripper: gripper_controller/gripper_cmd (control_msgs/action/GripperCommand)

Standard ROS2 gripper action. Position 0.0 (closed) to 1.57 (open).

ros2 action send_goal /gripper_controller/gripper_cmd control_msgs/action/GripperCommand \
  "{command: {position: 0.0, max_effort: 100.0}}"

Topics

Topic	Type	Description
pink_ik/current_ee_pose	geometry_msgs/PoseStamped	Current FK end-effector pose at 10 Hz
pink_ik/status	std_msgs/String	Solver status: idle, solving, executing, reached, failed
prometheus/joint_states	sensor_msgs/JointState	All joint positions (arm + gripper)

Configuration

IK parameters are in config/pink_ik_params.yaml:

• position_cost / orientation_cost — task weights (position prioritized for 5-DOF arm)
• max_iterations / dt — differential IK loop settings
• position_threshold — convergence criterion (meters)
• trajectory_duration / num_waypoints — trajectory execution timing
• named_poses — dict of named joint configurations (home, ready, tucked)

Simulation and Visualization

1. Gz Sim (small_house World):

2. Rviz (small_house World):

Acknowledgment

this project is heavily based on the follwing repos

• BCR_BOT
• SO-ARM-100