SiMMER (Legged Mobile Manipulator)

SiMMER is a ROS 1 (catkin) project for a 6-legged mobile base with a manipulator. It contains two catkin packages:

• lmm_model: robot description (URDF), meshes, Gazebo controllers, and simulation launch files
• lmm_sc: supervisory control stack for real-time planning, joystick input handling, gait planning, redundancy resolution, and hardware servo output

Project background (from the original documentation):

• Version: 1.1
• Date: 26 February 2020
• Core ideas: Adaptive Stroke Length, Full Cycle Planning, optimization-based solving (SciPy fsolve)

---

Repository Structure

SiMMER/
├── README.md
├── lmm_model/
│   ├── urdf/lmm.urdf
│   ├── meshes/*.STL
│   ├── config/controllers.yaml
│   ├── launch/{display.launch,gazebo.launch}
│   └── src/joint_controller.py
└── lmm_sc/
    ├── launch/{start.launch,simulation.launch}
    ├── config/joints.yaml
    └── src/
        ├── master.py
        ├── joy_incremental.py
        ├── motor_controller.py
        ├── tip_tf_broadcaster.py
        └── modules/*.py

---

Requirements

System

• Ubuntu with ROS 1 (catkin-based setup)
• Python for ROS rospy nodes

ROS packages used in this repository

• catkin
• roscpp
• rospy
• sensor_msgs
• std_msgs
• robot_state_publisher
• joint_state_publisher
• rviz
• gazebo / gazebo_ros
• controller_manager
• joy
• i2cpwm_board (required for hardware motor output path)

Python modules used by control nodes

• numpy
• scipy (for optimization-based solver paths)
• rospkg
• yaml
• ROS TF libraries (tf2_ros, TF messages)

---

Setup (Catkin Workspace)

1. Create a catkin workspace (if you do not already have one):
   • mkdir -p ~/catkin_ws/src
   • cd ~/catkin_ws/src
2. Place this repository in ~/catkin_ws/src/SiMMER.
3. Build:
   • cd ~/catkin_ws
   • catkin_make
4. Source:
   • source devel/setup.bash

Note: In this repository snapshot, catkin_make availability depends on your ROS installation and shell environment.

---

Running the Project

1) Visualization (URDF + RViz)

Run:

• roslaunch lmm_model display.launch

This starts:

• joint_state_publisher
• robot_state_publisher
• rviz with lmm_model/rviz/lmm.rviz

2) Gazebo Simulation + Supervisory Control

Run:

• roslaunch lmm_sc simulation.launch

This launches Gazebo, spawns the model, loads controllers, and starts:

• joystick input node
• TF tip broadcaster
• supervisory controller (master.py)
• simulation joint command bridge (lmm_model/src/joint_controller.py)

3) Hardware-oriented supervisory control path

Run:

• roslaunch lmm_sc start.launch

This runs:

• joystick input
• supervisory controller
• motor controller (motor_controller.py) publishing ServoArray to servos_absolute
• i2cpwm_board node for PWM/servo output

---

Main Nodes and Responsibilities

lmm_sc/src/master.py

Central real-time supervisory control node:

• subscribes to incremental end-effector commands (lmm_incremental_inputs)
• performs redundancy handling and inverse kinematics
• plans trunk-body and leg-tip trajectories when base motion is required
• publishes complete robot joint states (lmm_joint_states)
• logs runtime data to lmm_sc/data/<timestamp>.csv

lmm_sc/src/joy_incremental.py

• subscribes to joy (sensor_msgs/Joy)
• maps joystick axes to Cartesian incremental commands
• publishes lmm_incremental_inputs (Float32MultiArray)

lmm_sc/src/motor_controller.py

• subscribes to lmm_joint_states
• maps joint angles to servo board input values using lmm_sc/config/joints.yaml
• publishes servos_absolute (i2cpwm_board/ServoArray)

lmm_model/src/joint_controller.py

• simulation bridge node
• subscribes to lmm_joint_states
• republishes each joint command to Gazebo controllers under /lmm/*_position_controller/command

lmm_sc/src/tip_tf_broadcaster.py

• publishes tip transforms for all six legs and the end-effector tip on /tf

---

Core Topics

• joy (sensor_msgs/Joy) -> joystick input
• lmm_incremental_inputs (std_msgs/Float32MultiArray) -> Cartesian increments
• lmm_joint_states (sensor_msgs/JointState) -> planned joint commands/state
• /lmm/.../command (std_msgs/Float64) -> Gazebo controller commands
• servos_absolute (i2cpwm_board/ServoArray) -> hardware servo outputs

---

Key Configuration Files

• lmm_sc/src/modules/inputs.py

  • kinematic dimensions
  • initial trunk/end-effector pose
  • timing (f_in, T_in, T_stroke)
  • manipulability thresholds and optimization bounds

• lmm_sc/config/joints.yaml

  • per-joint servo calibration:
    • servo channel
    • min/max joint angles
    • min/max board input values
    • axis direction multiplier

• lmm_model/config/controllers.yaml

  • Gazebo effort position controllers and PID gains

---

Data Logging

master.py automatically logs runtime data rows to:

• lmm_sc/data/<YYYY_MM_DD__HH_MM_SS>.csv

Logged values include time, input increments, end-effector and trunk position terms, manipulability indicator, stroke length, and selected joint angles.

---

Notes

• This project is ROS 1/catkin oriented.
• Some package metadata is still template/default (for example, lmm_sc/package.xml license field is TODO).
• The repository contains both simulation and hardware pathways; choose launch files accordingly.