• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.3
• /
• pp.248-253
• /
• 2015

In order to adapt to human's life and perform missions, a humanoid robot needs a height at least similar with children's. In this paper, we proposed a humanoid robot which is like a child who is taller than 1m. We presented showing the humanoid robot's kinematics, designing of a three-dimensional model, developing mechanisms, and the hardware structures using servo motors and compact size PC. Through this process, we designed and manufactured child humanoid robot 'CHARLES(Cognitive Humanoid Autonomous Robot with Learning and Evolutionary Systems)' that is robot is 1m 10cm tall and 8.16kg in weight. For robot's walking, we applied to ZMP-based walking technique and the creation algorithm is applied for walking patterns. Through experiments, we analyzed walking patterns according to the creation and changing parameter values.

• Journal of The Korean Association of Information Education
• /
• v.18 no.1
• /
• pp.25-33
• /
• 2014

Due to the recent development of educational robot hardwares, processing speed and scalability have been greatly improved. Thus, the robot hardwares that are compatible with temperature sensor for MBL and gyro sensor made a data logging possible. Students can conduct an experiment on scientific research and prediction, collecting and data analysis with robots that can process data logging. Therefore this research constructed and adopted science project class that introduced a Smart r-Learning that utilizes Class SNS and smartphone. As a result of applying a data logging smart r-Learning to elementary school 5th graders, it has shown that the students' logical thinking ability four of the six areas have been improved in t-test.

• The Journal of the Korea Contents Association
• /
• v.19 no.12
• /
• pp.164-172
• /
• 2019

In this paper, we implement a wheeled mobile robot that accurately and autonomously finds the optimal route from the starting point to the destination point based on computer vision in a complex indoor environment. We get a number of waypoints from the starting point to get the best route to the target through deep reinforcement learning. However, in the case of autonomous driving, the majority of cases do not reach their destination accurately due to external factors such as surface curvature and foreign objects. Therefore, we propose an algorithm to deepen the waypoints and destinations included in the planned route and then correct the route through the waypoint recognition while driving to reach the planned destination. We built an autonomous wheeled mobile robot controlled by Arduino and equipped with Raspberry Pi and Pycamera and tested the planned route in the indoor environment using the proposed algorithm through real-time linkage with the server in the OSX environment.

• The Journal of Korea Robotics Society
• /
• v.10 no.2
• /
• pp.104-111
• /
• 2015

In this paper, we proposed visual information to provide a highly maneuverable system for a tele-operator. The visual information image is bird's eye view from UFR(Unmanned Flying Robot) shows around UGR(Unmanned Ground Robot). We need UGV detection and tracking method for UFR following UGR always. The proposed system uses TLD(Tracking Learning Detection) method to rapidly and robustly estimate the motion of the new detected UGR between consecutive frames. The TLD system trains an on-line UGR detector for the tracked UGR. The proposed system uses the extended Kalman filter in order to enhance the performance of the tracker. As a result, we provided the tele-operator with the visual information for convenient control.

• International Journal of Advanced Culture Technology
• /
• v.11 no.1
• /
• pp.312-319
• /
• 2023

This study aims to enhance the engineering education for non-majors by incorporating the concept of Global Citizenship Education and addressing the need for education that responds to climate and ecological changes. The study uses robot programming as a tool to foster the development of global citizens. Non-majors often struggle with producing more than just motionless forms or solid productions, due to a lack of understanding of mechanisms and coding. The study proposes the use of the Convergence D-SteamRobot (DSR) to address this issue by blending humanities and engineering. This is achieved by presenting problems through books to increase empathy, integrating simple machine mechanisms, and creating prototypes to solve self-defined problems. Through this process, learners determine the SDGs topic they want to solve and learn about the simple mechanical mechanism involved in producing the prototype. The educational model provides a constructivist learning environment that emphasizes empathy and exploration, encourages peer-learning, and improves divergent thinking and problem-solving skills.

• The Journal of Korea Robotics Society
• /
• v.15 no.4
• /
• pp.398-403
• /
• 2020

Robot's throwing control is difficult to accurately calculate because of air resistance and rotational inertia, etc. This complexity can be solved by using machine learning. Reinforcement learning using reward function puts limit on adapting to new environment for robots. Therefore, this paper applied deep reinforcement learning using neural network without reward function. Throwing is evaluated as a success or failure. AI network learns by taking the target position and control policy as input and yielding the evaluation as output. Then, the task is carried out by predicting the success probability according to the target location and control policy and searching the policy with the highest probability. Repeating this task can result in performance improvements as data accumulates. And this model can even predict tasks that were not previously attempted which means it is an universally applicable learning model for any new environment. According to the data results from 520 experiments, this learning model guarantees 75% success rate.

• The Journal of Korea Robotics Society
• /
• v.19 no.1
• /
• pp.92-97
• /
• 2024

This paper provides a study of distributional perspective on reinforcement learning for application in mobile robot navigation. Mapless navigation algorithms based on deep reinforcement learning are proven to promising performance and high applicability. The trial-and-error simulations in virtual environments are encouraged to implement autonomous navigation due to expensive real-life interactions. Nevertheless, applying the deep reinforcement learning model in real tasks is challenging due to dissimilar data collection between virtual simulation and the physical world, leading to high-risk manners and high collision rate. In this paper, we present distributional reinforcement learning architecture for mapless navigation of mobile robot that adapt the uncertainty of environmental change. The experimental results indicate the superior performance of distributional soft actor critic compared to conventional methods.

• Journal of Drive and Control
• /
• v.20 no.2
• /
• pp.15-23
• /
• 2023

In this study, two-dimensional location of crops for auto weeding was detected using deep learning. To construct a dataset for soybean detection, an image-capturing system was developed using a mono camera and single-board computer and the system was mounted on a weeding robot to collect soybean images. A dataset was constructed by extracting RoI (region of interest) from the raw image and each sample was labeled with soybean and the background for classification learning. The deep learning model consisted of four convolutional layers and was trained with a weakly supervised learning method that can provide object localization only using image-level labeling. Localization of the soybean area can be visualized via CAM and the two-dimensional position of the soybean was estimated by clustering the pixels associated with the soybean area and transforming the pixel coordinates to world coordinates. The actual position, which is determined manually as pixel coordinates in the image was evaluated and performances were 6.6(X-axis), 5.1(Y-axis) and 1.2(X-axis), 2.2(Y-axis) for MSE and RMSE about world coordinates, respectively. From the results, we confirmed that the center position of the soybean area derived through deep learning was sufficient for use in automatic weeding systems.

• The Journal of Korea Robotics Society
• /
• v.18 no.2
• /
• pp.143-154
• /
• 2023

In this study, a fast motion planning method for the swing motion of a 6x6 wheel-legged robot to traverse large obstacles and gaps is proposed. The motion planning method presented in the previous paper, which was based on trajectory optimization, took up to tens of seconds and was limited to two-dimensional, structured vertical obstacles and trenches. A deep neural network based on one-dimensional Convolutional Neural Network (CNN) is introduced to generate keyframes, which are then used to represent smooth reference commands for the six leg angles along the robot's path. The network is initially trained using the behavioral cloning method with a dataset gathered from previous simulation results of the trajectory optimization. Its performance is then improved through reinforcement learning, using a one-step REINFORCE algorithm. The trained model has increased the speed of motion planning by up to 820 times and improved the success rates of obstacle crossing under harsh conditions, such as low friction and high roughness.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.39 no.4
• /
• pp.393-402
• /
• 1990

An iterative learning contol method is presented for a class of linear periodic systems, in which a parameter estimator of the system together with an inverse system model is utilized to generate the control signal at each iteration. A convergence proof is given and two numerical examples are illustrated to show the validities of the algorithm. In particular, it is shown that the method is useful for the continuous path control of robot manipulators.