Back

Dot-to-Dot

Field

Robotics

Development Duration

July 29, 2019 - August 2, 2019

Components

AR Marker

Camera

Mobile Robot

Development Tools

openRTM-aist

Project type

Team Project

Background

I participated in RT Middleware Summer Camp 2019, where our team developed a robotic system as a group project.

What Is RT Middleware?

RT Middleware is a software platform for building robotic systems, known as RT Systems, by connecting multiple reusable software modules called RT Components.

The National Institute of Advanced Industrial Science and Technology, AIST, develops OpenRTM-aist and organizes the annual RT Middleware Summer Camp to train new users of the platform.

During the 2019 summer camp, participants worked in teams of four to develop a robotic system using RT Middleware. Over five days, each team completed the entire development process, from proposing a system concept to implementation and the final presentation.

OpenRTM-aist

Proposed System

Our system was named Dot-to-Dot.

The project recreated the traditional dot-to-dot puzzle, in which numbered points are connected in sequence to reveal an image, using a mobile robot.

System Configuration

The system consisted of multiple AR marker blocks and one differential-drive mobile robot.

Each AR marker block was a physical object with an AR marker printed on all four sides. All four markers had the same ID, making the block easier for the robot to recognize regardless of its viewing direction.

The mobile robot carried the following devices:

A PC for marker recognition and the drawing application
A PC for environmental sensing and autonomous navigation
An RGB camera
A laser range finder, URG

The marker-recognition PC and RGB camera detected the AR marker blocks, recorded their coordinates, and plotted them on a graphical user interface.

The environmental-sensing PC and laser range finder controlled the autonomous movement of the robot.

Operating Environment

The AR marker blocks were placed on the floor, and the robot traveled around them along the route indicated by the blue arrows.

Using data from the laser range finder, the robot continuously maintained a fixed distance from the wall. When it detected walls on two adjacent sides, it rotated 90 degrees in place to follow the next wall.

This behavior ensured that the RGB camera remained pointed toward the AR marker blocks throughout the scanning process.

RT Component Architecture

The system consisted of eight RT Components.

The components responsible for autonomous robot navigation using the laser range finder were grouped as the Robot RTC Group.

The ArUco Marker RTC Group handled AR marker detection.

The Integration RTC managed the marker coordinates and drew the connecting lines between the detected points.

Software Process

The RGB camera mounted on the mobile robot continuously searched for AR markers.

When a marker was detected near the center of the camera image, the system calculated its position relative to the robot. This information was then used to plot the marker’s coordinates on the GUI.

After the scanning process was complete, the system connected all recorded coordinates with straight lines to reveal the final image.

I served as the team leader and was also responsible for developing moveCommander2, a C++ RT Component that processed laser range finder data and controlled the robot so that it traveled parallel to the wall.

Test Results

We tested the system by arranging the marker blocks in the shapes of a star and a fish.

For the star-shaped arrangement, the system successfully acquired the marker coordinates and produced a clear drawing.

For the fish-shaped arrangement, the robot detected all of the markers. However, some points were plotted twice, and false detections caused points to appear in locations where no markers had been placed.