Description

Alternative method to generate odometry data for indoor robots using optical flow sensor.

Note: This branch tests the ideology by utilizing RGB camera and performing cross corelation algorithm in gazebo simulation environment.

Devices that can be used as optical flow sensors

• USB mouse - requires USB driver
• Optical Flow sensor module, e.g MATEKSYS 3901-L0X, Avago ADNS-9500 - requires low level communication driver as well as has some minimum height issues.
  • PWM3901
  • PAA5100JE
  • Libs for above sensor
• General Purpose Camera, e.g RaspberryPi camera - requires to design a custom cross-corelation algorithm.

Challenges:

Performance highly depends on the surface of the ground. Rough terrains, high reflective surfaces are not suitable.

Node Structure:

• Low level driver: Communicates with hardware : publishes data on individual sensor topic - x2 for each sensor
• High level driver: which publishes data on topic odom by subscribing to individual sensor topics and performing the mathematical operations.

Parameters:

• Position of sensor from center of robot - x, y, yaw

Topics:

• /optical_flow_sensor_a
• /optical_flow_sensor_b
• /odom

Services:

• Reset initial pose

Frames:

• odom => base_link

⚠️ Note: The initial position fo the robot is consider to be that from when the launch file is launched.

Mathematical Model

Note: Naming Convention: RGB => x, y, z

Positive direction for rotation follows right hand thumb.

General Math Model

(R_x)_M1: Translation in x-direction of Robot interpreted by Sensor M₁
(R_y)_M1 : Translation in y-direction of Robot interpreted by Sensor M₁
(R_ψ)_M1 : Rotation about z-axis of Robot interpreted by Sensor M₁

θ, (M₁)_θ : Orientational offset of sensor M₁ wrt Robot
(M₁)_px : Positional offset of sensor M₁ wrt Robot in x-direction
(M₁)_py : Positional offset of sensor M₁ wrt Robot in y-direction

R_x : Translation in x-direction of Robot
R_y : Translation in y-direction of Robot
R_ψ : Rotation about z-axis of Robot

Special Case

Tested On

• Ubuntu 22.04
• ROS2 Humble
• Gazebo Classic V11

Usage

To test the system in simulation environment,

Clone repository in your workspace

git clone git@github.com:maker-ATOM/optical-odometry.git

and perform,

colcon build --packages-select optical_odometry_gazebo --symlink-install

in the root of your workspace.

Launch the bringup file which launches rviz2, gazebo and spawns the turtlebot3.

ros2 launch optical_odometry_gazebo bringup.launch.py

ToDos

Issues

Handy Commands