Sungkeun Yoo

This study presents a model-based feedforward method for reducing velocity errors during the vertical travel of a magnetically coupled wall-climbing robot. Permanent magnets on opposite sides of the wall maintain magnetic coupling, while a spring-preloaded linkage adjusts the drive-wheel normal force. A dedicated test bench was used to measure the traction–slip characteristics of the urethane roller–acrylic contact at normal forces of 20 and 30 N. Magic Formula models were fitted to the measurements. Representative slip magnitudes were calculated from the models and the ascent force balance using a nominal equivalent resistance of the robot. These fixed magnitudes were applied to both ascent and descent through direction-dependent wheel-velocity commands. In experiments without longitudinal velocity feedback, the compensation reduced the absolute errors in mean velocity and travel distance, averaged over all tested conditions, by approximately 76% and 69%, respectively. The method was also integrated with proportional–integral velocity feedback and pure pursuit path tracking and assessed in repeated ascent–descent trials. The results demonstrate the error-reduction capability of the fixed feedforward compensation under the tested conditions.

In this paper, we propose simDP, a sim-to-real transfer framework that enables diffusion policies trained in simulations for the efficient deployment on real-world robots. The key idea is to reduce the sim-to-real gap by aligning the action and observation spaces between simulation and reality. Specifically, we reformulate the action space using end-effector pose and binary gripper state, which can be shared between simulated and physical robots. In addition, we use camera-based visual observations as the primary sensing modality in both domains and train a real-world observation encoder to align with the latent representation learned in simulation. This design allows the action decoder trained in simulation to be reused in the real-world with minimal modification. We evaluated simDP on object manipulation tasks derived from the MimicGen benchmark and show that a simulation-trained diffusion decoder, when combined with a real-world adapted observation encoder, achieves task completion performance similar to and in some cases better than diffusion policies trained only on limited real-world data. These results, obtained across four manipulation tasks in a calibrated real-world transfer setting, suggest that reusing a simulation-trained action decoder with lightweight real-world encoder adaptation provides an effective strategy for controlled sim-to-real transfer, while broader evaluation across diverse tasks, environments, and robot embodiments remains an important direction for future work.

This study proposes a locomotion method applicable to a 6-legged mobile robot with passive wheels. To achieve the desired movements of the robot, it is necessary to control the end points of the robot's legs to reach specific target positions. Leg end point positions can be defined relative to the robot's central coordinate system through kinematic modeling, which includes both forward and inverse kinematics. This study equips the proposed robot with passive wheels at the ends of its legs to facilitate movement through its actions. The robot's locomotion is planned by tracking an elliptical trajectory at its leg end points, and it gains propulsion for linear movement through the relationship between the absolute velocity of the wheels relative to the ground and the relative velocity of the legs in the robot's frame. The robot's performance is evaluated through 236 experiments, using the height, xaxis length, and y-axis length of the elliptical trajectory as design parameters. The results indicate that the optimal trajectory is achieved at a height of 92.5mm, with x-axis and y-axis lengths of 120mm and 70mm, respectively. Future research will focus on designing a mechanism for flexible switching between wheels and legs, as well as analyzing steering performance.

This article presents the design and position-tracking control of a novel two-degree-of-freedom dual-rope winch robot. It is designed to be facilely installed at both corners of a building using two synthetic ropes and to move freely on the wall plane using the ropes to perform building maintenance work. This robot consists of a dual-rope winch system responsible for its movement on a plane and a thrust module that allows the robot to remain close to the wall through the force of the propeller. In the dual-rope winch system, ascending modules driven by two pulleys and connected by differential gears are arranged symmetrically; the position and posture on the plane can be determined by controlling the ropes. The robot winds a synthetic rope around the pulley to support the load with frictional force. Precise position control is difficult to implement because of the elasticity of the rope and the slip around the pulley. Therefore, in this article, a feedforward control to predict and compensate for the slip length of the rope and a proportional-integral controller to regulate the position error owing to the dynamic characteristics of the rope are proposed. Experiments to verify the performance of the controller for various trajectories were performed on the test bench. It was confirmed that the actual trajectory deviation is within 0.5% of the target trajectory.

In recent years, cable-driven-parallel robots (CDPRs) have been studied for façade operations. There are various types of CDPRs; however, underconstrained CDPRs are capable of wider operating in façade workspaces than overconstrained CDPRs. Therefore, in this article, a dual ascender robot (DAR) was used for façade operations. Herein, two suggestions for safe façade operations are presented. First, a flexible nylon fiber rope was modeled such that the vibration direction, natural frequency, and damping ratio of the DAR could be converted through a Jacobian matrix and modal decomposition from the rope model. Second, input shaping control was applied to reduce vibrations, based on the vibration model of a DAR using the rope model. Modal decomposition was verified using a verification experiment, and the effect of input shaping was evaluated by comparing the w/input shaping and w/o input shaping experiments. w/input shaping case was shown about 48% reducing robot vibration and about 35% shortening settling time compare with w/o input shaping case.

This paper presents a rope winch with high position repeatability using a differential gear mechanism. This winch is easy to apply to existing buildings using synthetic fiber ropes, which are generally used by building exterior maintenance workers, and can be widely utilized as a winch of a gondola or climbing robot. Multiple pulleys are used to achieve high payloads, and differential gear mechanisms are used to distribute the torque across each pulley evenly. Uniform torque distribution minimizes slipping on each pulley, which reduces the power loss due to friction. This is proven through an analysis using the capstan equation and an experiment on a test bench. In addition, a criterion for the position repeatability of the winch is designed, and its results show that one pressure roller achieves optimal performance when the roller at the free end is pressed to 37 N. Extensive experiments were carried out on the final model of the winch, and position repeatability of under 50 mm deviation per 10 m was achieved through stable torque distribution under various payloads.