The Technology behind 'Kokko MK' and Angchickin's impressions

The Engineering Education Innovation Center club 'Angchickin' participated in the [2023 Defense Robot Contest] hosted by the Ministry of National Defense and the Korea Institute of Robotics and Convergence Research. The team won the top prize in the Defense Intelligent Robot category (Minister of National Defense Award).  Kang Minji, the president of Angchickin 2023, along with participants No Si Tae and Choi Ji Won, participated in an interview to talk about the prize.

The club 'Angchickin' was established in 2019 as a robotics academic club at Hongik University's Sejong Campus.  Since the club’s establishment, it has been participating in defense-related competitions such as the Defense Robot Contest and RoboCup Rescue (an annual international robotics engineering competition founded in 1996).  The club competes in various events including drone competitions, autonomous driving contests, and capstone projects, enhancing their capabilities and proving of their achievements through the winning of awards.

In the last 'Defence Robot Competition', the name of the product that Angchikin entered is called 'Kokko MK'.  Kokko MK is an autonomous unmanned robot designed to perform various tasks for the military.  It’s performance capabilities are as follows:

Kokko MK's high-powered motors and sturdy aluminium frame structure can easily hold two adult men.  The robot's modular design allows for the easy attachment and detachment of various devices and actuators depending on the situation.  A wide range of camera sensors enable stable autonomous driving, remote control, and low-latency real-time monitoring.  It can accurately recognize and process learned objects by applying image recognition technology using Deep Learning.

'Kokko MK' was built using a variety of technologies, and is divided into both software and hardware.

First, the software is divided into the sections: object recognition judgement, autonomous driving algorithms, middleware, and operating system.

Object recognition and discriminative technologies have been utilized in the following three ways.

YOLO (You Only Look Once) is a real-time object detection algorithm that is used to identify and localize various objects in images or videos.  It broadly encompasses technologies such as multi-labeled and key point, which have emerged as recent advances in the field of computer vision.  Among these, the YOLO model has a faster and more accurate data processing speed than the existing models.   In Kokko MK, YOLO is employed to recognize objects in the image input from the camera and perform image processing through OpenCV.

Open CV is an open-source library that can perform a variety of tasks related to computer vision.  For example, it is used for product auditing in factories, medical image processing, revision, and assessment, CCTV footage analysis, robotics, and more.  Kokko MK uses OpenCV to perform image processing, object tracking, and feature detection, amongst others.

Aruco Markers are artificially generated markers that are used by Kokko MK to track location and determine orientation. Each marker has a unique pattern of N x N dimensions, similar to a QR code, and each marker has a coordinate system enabling computers to discern the position of objects and their 3D orientation (x-axis, y-axis, z-axis).  This capability helps Kokko MK accurately ascertain position and direction.

The autonomous driving algorithm technology was utilized in the following three ways.

The Inertial Measurement Unit (IMU) combines sensors such as accelerometers and gyroscopes to measure the movement and orientation of Kokko MK.  It is a sensor that acts like the cochlea in the human ear.  Data from the IMU is used to assess the driving status  Kokko MK, and run autonomous driving algorithms based on this information.

RGB frames are visual data obtained from a standard camera.  Utilizing these, Kokko MK recognizes the driving environment in real-time, detecting tracks, fog, and other elements to make appropriate driving decisions.

Depth frames are image data that includes distance information in 3D space.  Using depth frames, Kokko MK can assess distances to surrounding objects and make decisions about driving in various environments such as sand, gravel, water, and more.

Middleware is software that allows different applications to communicate with each other. Kokko MK uses DDS (Data Distribution Service) as its middleware.  DDS is a real-time data distribution middleware set as an international standard middleware by the OMG.  By utilizing DDS, Kokko MK can send and receive data to and from various sensors and devices in real time.  This enables real-time control of Kokko MK in external environments.

Finally, ROS2 (Robot Operating System 2) is a robot operating system platform used for developing robot software.  ROS2 provides the ability to efficiently manage and run various robotic systems.  Using ROS2, Kokko MK is able to seamlessly integrate various technology and hardware components to execute programs and send and receive data in real-time. 

The following describes the hardware aspects used in the development of Kokko MK.

The overall design of the robot was created using Autodesk Intentor and Fusion360.  While Kokko MK was designed with modules to develop a robot capable of adapting to various environments, the structure of the robot can be changed according to specific requirements.

Additionally, it was designed to facilitate the detachment of additional robot mechanisms or actuators as needed.

Kokko MK’s specifications are as follows.

Mass: 32kg Materials: Aluminum profiles, aluminum plates, steel plates 

Operating time: Average 3 hours 

Advantages: Able to securely tow objects up to 200kg 

Features: Money pulley able to securely grip objects weighing up to 2kg

The club 'Angchikin' first attempted a large-scale robot development project in 2021.  The club experienced difficulties due to insufficient choice and concentration, resulting in the robot malfunctioning during the competition. They then realized the importance of "communication" and "time management" with team members.  During the second competition, the 2023 Defense Robotics Competition, this previous experience helped them to place importance on constant communication with team members.  The club said that they increased the frequency of meetings for communication, shared goals on the whiteboard, and daily updated team members with the latest developments.  Through this, members were able to clearly understand the team's goals and progress.  In addition, "prioritizing" and "selecting and focusing" on development goals, which was not possible in the first competition, is how the team was able to win the Secretary of Defense Award this time around.  In addition, by placing importance on "prioritizing" and "selection and focus" in relation to development goals, is how the team won the Defense Minister's Award this time around.

In addition, for students who need to build something for the first time, the most basic thing they need to know is an understanding of the skills and tools needed for their product.  For example, software development requires an understanding of programming languages, frameworks, and libraries, while hardware development requires an understanding of electronics, circuit design, and sensors, amongst others.  Another important factor is to be able to solve problems that arise during the development of a product. One needs to be able to analyze problems, find appropriate approaches, and implement effective solutions.

Online Communications Reporter Seo Jihwa