Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Dongil Choi has a total of 32 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Deep Reinforcement Learning Control of a Hexapod Robot

    by Taesoo Kim, Minjun Choi, Seunguk Choi, Taeuan Yoon, Dongil Choi

    Recent advances in legged robotics have highlighted deep reinforcement learning (DRL)-based controllers for their robust adaptability to diverse, unstructured environments. While position-based DRL controllers achieve high tracking accuracy, they offer limited disturbance rejection, which degrades walking stability; torque-based DRL controllers can mitigate this issue but typically require extensive time and trial-and-error to converge. To address these challenges, we propose a Real-Time Motion Generator (RTMG). At each time step, RTMG kinematically synthesizes end-effector trajectories from target translational and angular velocities (yaw rate) and step length, then maps them to joint angles via inverse kinematics to produce imitation data. The RL agent uses this imitation data as a torque bias, which is gradually annealed during training to enable fully autonomous behavior. We further combine the RTMG-generated imitation data with a decaying action priors scheme to ensure both initial stability and motion diversity. The proposed training pipeline, implemented in NVIDIA Isaac Gym with Proximal Policy Optimization (PPO), reliably converges to the target gait pattern. The trained controller is Tensor RT-optimized and runs at 50 Hz on a Jetson Nano; relative to a position-based baseline, torso oscillation is reduced by 24.88% in simulation and 21.24% on hardware, demonstrating the effectiveness of the approach.


  • Open Access

    Monocular Vision-Based Obstacle Height Estimation for Mobile Robot

    by Seongmin Ahn, Yunjin Kyung, Seunguk Choi, Dongyoung Choi, Dongil Choi

    For a robot to operate robustly in diverse real-world environments, reliable obstacle perception is essential, which fundamentally requires depth information of the surrounding scene. Monocular depth estimation provides a lightweight alternative to active sensors by predicting depth from a single RGB image. However, due to the absence of sufficient geometric and optical cues, it suffers from inherent depth ambiguity. To address this limitation, we propose R-Depth Net, a monocular absolute depth estimation network that utilizes distance-dependent defocus blur variations and optical flow as complementary depth signals. Furthermore, based on the depth maps generated by R-Depth Net, we design an algorithm for obstacle height estimation and traversability assessment. Experimental results in real-world environments show that the proposed method achieves an average RMSE of 0.30 m (15.7%) and MAE of 0.26 m (15.7%) for distance estimation within the 1.0–3.0 m range. For obstacle height estimation in the range of 0.10–0.20 m, the system achieves an average RMSE of 0.048 m (29.3%) and MAE of 0.040 m (26.4%). Finally, real-time deployment on a quadruped robot demonstrates that the estimated depth and height are sufficiently accurate to support on-board obstacle traversal decision-making.


  • Open Access

    Development and Balancing Control of Control Moment Gyroscope (CMG) Unicycle–Legged Robot

    by Seungchul Shin, Minjun Choi, Seongmin Ahn, Seongyong Hur, David Kim, Dongil Choi

    A wheeled–legged robot has the advantage of stable and agile movement on flat ground and an excellent ability to overcome obstacles. However, when faced with a narrow footprint, there is a limit to its ability to move. We developed the control moment gyroscope (CMG) unicycle–legged robot to solve this problem. A scissored pair of CMGs was applied to control the roll balance, and the pitch balance was modeled as a double-inverted pendulum. We performed Linear Quadratic Regulator (LQR) control and model predictive control (MPC) in a system in which the control systems in the roll and pitch directions were separated. We also devised a method for controlling the rotation of the robot in the yaw direction using torque generated by the CMG, and the performance of these controllers was verified in the Gazebo simulator. In addition, forward driving control was performed to verify mobility, which is the main advantage of the wheeled–legged robot; it was confirmed that this control enabled the robot to pass through a narrow space of 0.15 m. Before implementing the verified controllers in the real world, we built a CMG test platform and confirmed that balancing control was maintained within ±1∘.


  • Open Access

    Monocular Vision-Based Obstacle Height Estimation for Mobile Robot

    by Seongmin Ahn, Yoonjin Kyung, Seunguk Choi, Dongyoung Choi, Dongil Choi

    To ensure that robots can operate reliably in diverse environments, obstacle detection is essential, which requires the acquisition of depth information of the surrounding environment. One approach for obtaining depth information is monocular depth estimation, a technique that predicts scene depth from a single camera input. However, monocular images inherently provide limited cues for depth perception, which restricts estimation accuracy. To address this limitation, we propose R-Depth Net, a monocular depth estimation network that leverages variations in blur intensity with respect to distance and frame-to-frame changes induced by motion as primary cues. Based on the depth maps generated by R-Depth Net, we developed an algorithm that enables a robot to estimate obstacle height and determine whether traversal is feasible. When applied to a quadruped robot platform, the proposed method achieved RMSE 0.30 m and MAE 0.26 m for obstacle distance estimation, and RMSE 0.048 m and MAE 0.040 m for obstacle height estimation. Furthermore, real-world obstacle traversal experiments demonstrated the effectiveness of the proposed monocular camera–based obstacle detection and traversal decision-making framework.


  • Open Access

    Control of a longitudinally extended two-wheeled inverted pendulum robot with a sliding mechanism

    by David Kim, Dongil Choi

    This paper proposes a balance control strategy for the longitudinally extended two-wheeled inverted pendulum robot (LE-TWIPR), which integrates wheel control with a sliding mechanism. By compensating for continuous center of mass shifts caused by disturbances and unbalanced payloads, the proposed method achieves stable control even under continuous external force conditions. The controller combines model predictive control with a disturbance observer that estimates external forces acting on the system in real time. Distance sensors mounted at the front and rear of the robot are used to estimate the slope angle, enabling collision avoidance on inclined surfaces. Based on the estimated external force and slope angle, slider compensation references are generated to maintain dynamic stability. Experimental results demonstrate that the LE-TWIPR can maintain balance and travel at a maximum speed of 1.5m/s under unbalanced payload conditions, and successfully traverse an $8^\circ$ slope. These findings validate the effectiveness and real-time capability of the proposed control approach.


  • Open Access

    Control of a Self-Balancing Two-Wheeled Cargo Transport Robot with Separation-Combination Capability

    by Seungchul Shin, Minjun Choi, Taesoo Kim, Seongyong Hur, David Kim, Dongil Choi

    The Single Inverted Pendulum (SIP) model is widely employed in various robotics applications for balance control due to its structural simplicity and ease of implementation. Self-balancing two-wheeled transport robots offer the advantage of maintaining stability while enabling dynamic acceleration and deceleration. However, their performance is limited when carrying heavy loads, and maintaining stable balance during the loading process remains challenging. To address these limitations, this study proposes a control strategy for a pair of self-balancing two-wheeled transport robots that operate cooperatively through a separation–combination mechanism. Simulations were conducted to compare the performance of the Linear Quadratic Regulator (LQR) and Model Predictive Control (MPC) under a self-balancing condition with an external force of 100 N applied. Additionally, the proposed separation–combination method was evaluated through a simulation involving the transfer of a 20 kg delivery module. The results showed that both methods could stably transport the load, while the MPC-based approach reduced the wheel torque by approximately 50% compared to the LQR method. These findings indicate that a two-wheeled robot employing MPC in conjunction with the separation–combination method is well-suited for transporting heavy loads.

  • Open Access

    Effect of Droplet Contamination on Camera Lens Surfaces: Degradation of Image Quality and Object Detection Performance

    by Hyunwoo Kim, Yoseph Yang, Youngkwang Kim, Dong-Won Jang, Dongil Choi, Kang Park, Sangkug Chung, Daegeun Kim

    Recently, camera sensors have been widely used in a variety of applications, including advanced driver assistance systems (ADASs), surveillance systems, and unmanned aerial vehicles (UAVs). These sensors are often integrated with intelligent algorithms to automatically analyze information and perform specific functions. However, during rainy weather, droplets on the camera lens surface can obstruct the view, leading to degraded image quality and reduced algorithm performance. This paper quantitatively evaluates the effect of droplet contamination on image quality and object detection performance. Image quality degradation was analyzed using the modulation transfer function (MTF), with droplet volume and number as variables. Results show that the MTF50 decreased by up to 80% when the droplet volume reached 10 μL. To assess the effect on object detection, performance changes were evaluated across different detection algorithms. The findings reveal that droplet contamination can reduce the detection performance of small objects by up to 90%, as measured by the mean average precision (mAP) metric. Furthermore, degradation was more severe on hydrophilic surfaces compared to hydrophobic ones. This study demonstrates that droplet characteristics such as volume, number, and shape significantly influence both image quality and object detection performance. It provides critical insights into selecting appropriate camera lens materials by comparing hydrophilic and hydrophobic surfaces while also highlighting the susceptibility of intelligent algorithms to environmental factors and underscoring the importance of effective cleaning techniques.


  • Control of a Two-wheeled Self-balancing Robot With a Sliding Mechanism

    by David Kim, Hyun Seok Kim, Dongil Choi

    Two-wheeled self-balancing robot maintains balance and performs driving functions using two wheels, featuring excellent mobility due to its simple structure and wheel-driven system. However, when a long longitudinal structure is applied to increase space efficiency, challenges may arise in balance control and slope entry. In this study, a sliding mechanism and Model Predictive Control (MPC) were applied to achieve balance and driving control for a robot with a long longitudinal body. The system dynamics were analyzed using the Lagrangian method, and a reference for the sliding mechanism was generated to support balance control. Through Gazebo simulations, the performance of the MPC algorithm was verified, confirming that the robot could reach a stable balance state through Swing-up motion and maintain balance consistently. Additionally, stable driving performance was demonstrated in slope driving tests. We conducted experiments by applying the same algorithm to the actual robot, confirming that it successfully performed stable balance control and slope driving control in real-world environments.


  • Large language model based collaborative robot system for daily task assistance

    by Seunguk Choi, David Kim, Myeonggyun Ahn, Dongil Choi