RoboOmni

A ROS 2 (Jazzy) mobile robotics platform built on a mecanum-wheeled chassis and used to collect datasets (RGB-D, LiDAR, IMU, wheel odometry). The system is engineered end-to-end: a custom ros2_control hardware interface talks to an ESP32 via micro-ROS, custom mecanum Jacobian controller plugins handle the inverse kinematics, and a single launch file brings up sensors, control, and visualization.

Demo

Robot

Visualization

Sensor outputs

LiDAR	Kinect	Kinect + LiDAR

Features

• Mecanum omnidirectional drive — 4 independently driven wheels for holonomic motion
• Custom control stack — ros2_control hardware interface + Jacobian / inverse-Jacobian controller plugins
• Embedded integration — ESP32 firmware bridged over micro-ROS (serial agent)
• Perception sensors — Kinect RGB-D camera, RPLIDAR, MPU9250 IMU
• Dataset-oriented design — synchronized wheel odometry, IMU, RGB-D, and 2D LiDAR streams for SLAM / perception research
• Single-command bringup — conditional launch args to enable/disable each subsystem

Hardware

Component	Detail
Chassis	Mecanum, 4 × 75 mm wheels
Motors	4 × DC motors with encoders, driven by voltage commands
MCU	ESP32 running micro-ROS firmware
Camera	Kinect RGB-D
LiDAR	RPLIDAR over /dev/ttyUSB0
IMU	MPU9250
Agent link	micro-ROS agent on /dev/ttyUSB1

Architecture

flowchart LR
    subgraph sim["mecanum_description"]
        URDF["robot.urdf.xacro (URDF model)"]
    end

    subgraph bringup["mecanum_bringup"]
        SYS["system_bringup.launch.py"]
        CTRL["controllers.yaml:<br/>jacobian · inverse_jacobian<br/>joint_state_broadcaster"]
        KIN["kinect_bringup (RGB-D)"]
        LID["lidar_bringup (RPLIDAR)"]
    end

    subgraph hw["mecanum_hardware_interface"]
        HWI["MicroRosInterface<br/>voltage cmd · position/velocity state · IMU"]
    end

    FW["ESP32 micro-ROS firmware<br/>(motor control · encoders · MPU9250)"]

    SYS --> URDF
    SYS --> CTRL
    SYS --> KIN
    SYS --> LID
    CTRL --> HWI
    HWI <-- "micro-ROS agent (serial)" --> FW

Loading

Wheel commands flow top-down (cmd_vel → Jacobian controller → per-wheel voltage → hardware interface → firmware), while odometry and IMU data flow bottom-up back to the controllers.

Repository ecosystem

This is the main workspace repo. The rest of the robot's software is split across dedicated repositories, all pinned in src/.repos:

Repository	Role
mecanum_jacobian_controller	ros2_control controller plugins: mecanum Jacobian + inverse-Jacobian (kinematics, TF output)
mecanum_hardware_interface	Custom ros2_control system interface: micro-ROS and UDP hardware implementations
mecanum_microros_firmware	ESP32 firmware: motor control, encoders, MPU9250, FreeRTOS tasks over micro-ROS
mecanum_udp_firmware	Alternate ESP32 firmware and UDP bridge variant

Workspace layout

src/
├── .repos                      # vcs manifest — imports external repositories
└── mecanum/
    ├── mecanum_description/    # URDF/xacro model, meshes, RViz config, visualization launches
    └── mecanum_bringup/        # system bringup launches + controllers.yaml

The workspace is a partial checkout: external repositories under src/control/, src/firmware/, and src/third_party/ are fetched with vcs (see below).

Quickstart

Requires Ubuntu 24.04 and ROS 2 Jazzy (source /opt/ros/jazzy/setup.bash).

# 1. Fetch external repositories (controllers, hardware interface, firmware, sensor drivers)
cd src && vcs import < .repos && cd ..

# 2. Install dependencies and build
rosdep update
rosdep install -i --from-paths src/ -y
colcon build

# 3. Run the full system
source install/setup.bash
ros2 launch mecanum_bringup system_bringup.launch.py

The two local packages are pure assets (launch/xacro/mesh/config) and also build standalone: colcon build --packages-select mecanum_description mecanum_bringup

Bringup options

system_bringup.launch.py accepts boolean flags to enable/disable subsystems:

Argument	Default	Purpose
use_kinect	true	Kinect RGB-D camera
use_lidar	true	RPLIDAR
use_micro	true	Microcontroller / micro-ROS agent
use_control	true	ros2_control system
use_display	false	RViz visualization

Visualization

ros2 launch mecanum_description display.launch.py

Micro-ROS agent (manual)

ros2 run micro_ros_setup create_agent_ws.sh src/third_party
colcon build

Sensors & frames

Sensor	Port / topic	Frame
LiDAR	/dev/ttyUSB0	lidar_link
micro-ROS agent	/dev/ttyUSB1	—
Kinect RGB	/kinect/rgb/...	kinect_rgb_optical_link
Kinect depth	/kinect/depth/...	kinect_depth_optical_link

Note: ROS_DOMAIN_ID=1 is set in the Dockerfile. Controller and interface naming (chassis_fr/fl/rl/rr_wheel_joint, voltage command interface) must stay in sync across robot_core.xacro, ros2_control.xacro, and config/controllers.yaml.

License

Apache License 2.0. See LICENSE.