Prototype of the system simulating the force acting on the object, referring to the force from the IMU Sensor value.

This repository focuses on developing firmware and implementing signal-processing filters for the STM32 G474RE microcontroller. The goal is to accurately measure and analyze three-axis inertial data (X, Y, and Z) to support simulation in gazebo.

Table of Contents

Objective

• To create a simulation system for the movement of objects when a force is applied to the Gazebo by using the actual force obtained by converting the value from the IMU Sensor into the force that will be applied to the object on the Gazebo.

Scope

• Actual Object

  • The actual object is the M5StickC Plus microcontroller.

• Simulation System

  • The system is constructed on Gazebo with the object model being a Square shape of the same size as the actual object.
  • The force applied to the object in the simulation system is only applied at the center of the object.
  • The force applied will only be in the x and y axes.

• Working environment

  • The operation will be performed under the A4 size test area with the Web camera set up 0.3 m above the paper.

  

User installation guide

Prerequisites

Ensure you have the following dependencies installed

Hardware Require

• M5 StickC Plus
• Web camera

Middle ware

• ROS2 Humble
• MicroRos

Applications

• Gazabo
• Platfrom IO

Install project workspace

Clone this workspace

git clone https://github.com/DenWaritthon/imu_force_ws.git -b main

Methodology

IMU filter : IIR filter

It is a filter used to adjust the acceleration value obtained from the IMU Sensor to be more stable by taking the previous value into consideration and adjusting the properties according to the specified gain value which will work according to the equation.

$$a_{out} = a_k a_{in} + (1-a_k)a_{out}[n-1]$$

Given

• $a_{out}$ is the acceleration out put from IIR filter
• $a_{in}$ is the acceleration input put from IMU sensor
• $a_k$ is gain value

Newton's equations

Convert the accelration value obtained from the IMU into a force to apply to the object in Gazabo using Newton's equations.

$$\sum F = ma$$

Given

• $F$ is the force to be obtained
• $m$ is the weight of the actual object, which weighs 0.02 kg.
• $a$ is the acceleration measured by the IMU

Aruco

Aruco functions to detect an area from an Aruco Marker by using its four corners to define a working area and detect black objects. Once the area of interest is identified, a point is created at the center of the area to serve as the origin (0,0), and the position of the black object within that area is then detected.

System architecture

System separate the work into sub-node that work differently 8 node consist of

• M5StickC_IMU_node (Micro Ros) is a node that have input value from IMU and Pub the value to Topic : /IMU_publisher to the node IMU_calibrate_node and lowpass_accel_collector_node.

• imu_calibration_node is a Node that have input value from IMU_node (Micro ROS) to Calibrate value of the IMU to make it more accurate.

• imu_filter_node is a Node that have a input value from M5StickC_IMU_node (Micro Ros) after Calibate the value then the accelerate value by Pub Topic : /acceleration to Node Force_control_node.

• aruco_detect_node is a Node that have input image from webcam for detect Black object (Now is a M5StickCPlus) to make localization from Aruco Area to Pub Topic : /object_pos to aruco_controller node.

• aruco_controller is a Node that have input from aruco_detect_node and gazebo_node to calculation difference position and Pub topic : /diff_pose to Force_control_node.

• Force_control_node is a Node that get input of acceleration value from Node : imu_filter_node to calculate in Newton's equations and get input of difference between actual position and position in Gazebo from Node : aruco_controller to control Object by Servive :/apply_link_wrench of Gazebo.

• Gazebo_node Node of Gazebo program that will be display result of the system that have Topic : /Odom for display the position of the Object on Ground_truth.

• model_position_node is a Node that have input value from Node : Gazebo_node to calculate model position.

User guide

After Clone workspace then Build and source

cd imu_force_ws/
colcon build
source install/setup.bash

Before use

To use need to build ChAruco first by go to src/aruco/scripts/python_file_Webcame/Gen_board.py will get ChArUco_Marker_A4.pdf file. Then print it out and take a picture by the file src/aruco/scripts/python_file_Webcame/cap.py to take a picture in a lots of angle and posture to Calibrate to make the camera accurate After take the picture use the file src/aruco/scripts/python_file_Webcame/Calibration.py to find the parameter to use in the next process

How to use simulation

Before starting the operation after setting up the environment.

• Run caliblate IMU

ros2 run imu_filter imu_calibrate.py

Run the simulation in ROS2.

• Run Gazebo simulation

ros2 launch object_simulation model_display.launch.py

• Run IMU filter

ros2 run imu_filter imu_filter.py

• Run Aruco

ros2 run aruco aruco_detect_roi_ros.py

• Run Force controller Always execute/perform last

ros2 run object_simulation force_controller.py

Check the position of objects in the Gazebo and the real world.

• In Gazebo

ros2 run object_simulation model_position.py 

• In real world

ros2 topic echo /object_pose

Demos and Result

Tutorial clip

Demo clip

Conclusion

From testing the system that use only IMU make the object not stop and move by itself but in the real world the object did’t move. So we use position from Aruco integrated with IMU to adjust the position of the object in Gazebo to make it similar to the real world. Firstly get a acceleration value from the IMU and convert it to force that do to the IMU to control the object in the first place. After the IMU don’t have any acceleration but the object was not at the target then use the Aruco to posion control to make the object in Gazebo was at the right target. The result of testing the object in gazebo move to the target but the system not consistent to the objective that use IMU as main sensor.

Future plan

• Explore advanced filtering algorithms (e.g., Kalman filters) for improved accuracy and robustness.
• Expand documentation and provide example code to facilitate integration into larger robotics or control systems.

Developer Member

• Waritthon Kongnoo
• Kraiwich Vichakhot
• Chayanin Napia