In recent years, educational-support robots have been attracting attention. In conventional robots, the leaner cannot learn in a environment of spontaneous learning where they decide the number of solved problems by themselves. As a remedy, we proposed the problem-number negotiation method by deciding the number of solved problems through a conversation with the learner before learning. Previous studies have suggested that the problem-number negotiation method improve the number of solved problems in short and long-term learning. However, we had not verified the impact of learning effect in the problem-number negotiation method. This paper examine the effect of long-term learning with a robot that uses the problem-number negotiation method (proposed robot) by university students. The experiment verified the learning effect compare with the non-conversational robot (conventional robot). Experimental result suggested that proposed robot improved the learning effect of learners compared to conventional robot.

This paper focuses on food samples for handling experiments, especially samples of pasta noodles. Challenges in the research and development of food handling robots include the difficulty of evaluating experiments under consistent conditions. The shape and mechanical properties of food items much affect their handling; however, these properties often vary substantially, making it difficult to standardize experimental conditions. So far, food samples have been applied to evaluation experiments, but most existing food samples focus on visual replication and do not reproduce deformation and surface properties that affect handling. Consequently, this paper proposes food samples designed for robot handling evaluation. The proposed food samples aim to replicate deformation and surface properties. Especially, we fabricated pasta noodle samples and applied the fabricated samples to handling experiments. First, pasta noodle samples were produced through molding. By comparing the deformation of cooked pasta noodles with that of pasta noodle samples, we selected liquid silicone materials capable of replicating the deformation of cooked pasta noodles. Additionally, liquid paraffin was applied to the surface of the pasta noodle samples to recreate the arrangement appearance. Then, gripping experiments were conducted using a needle gripper and a tong gripper to evaluate the effectiveness of the prototype pasta noodle samples. The results showed that the silicone pasta noodles made with Ecoflex 00-20 undercoated with liquid paraffin successfully replicated the properties of cooked pasta noodles with 0% salt.

This paper discusses the manipulation problem of tearing a flexible object and dividing it into target proportions by a single-axis tensile motion using a dual-arm robot with parallel grippers as end effectors. First, through preliminary experiments, we show that there are unstable phenomena such as initial value sensitivities and discontinuities in the tearing of the flexible object. We show that there is a characteristic relationship between the grasping ratio and the tearing ratio, which is discontinuous in some parts. Next, this relationship is expressed by a mathematical model called the tearing model. Based on the model, we propose a method to control the tearing ratio by 1-step or 2-step tearing motion. In the 2-step control, the volume distribution is monitored via the depth information , and the regrasping motion is conducted at the appropriate timing to repeat the tearing motion. Finally, we demonstrate the feasibility of the proposed method by experiments.

In this paper, we propose a Wire-driven Earthworm Robot that uses the peristaltic motion of worms to turn. The robot with a low height and small width is needed to inspect narrow spaces such as ceilings. We focused on the peristaltic motion of earthworms, which can move freely even in a narrow environment, and developed the robot that can turn and move by differentiating the tension of the left and right wires. As an evaluation experiment, a turning experiment using this method was conducted to determine the width required for turning.

Playing the violin is one of the important methods for expressing emotion for robots. In this study, we focused on the violin, and aimed to develop a system that automatically determines the bow movement from musical scores. We adopted Q-learning and ε-greedy method with a value function approximation by neural network. We adopted a different formula for calculating the rewards from our previous studies. We selected three musical pieces and several BPM values. As a result of the simulation, for all the musical pieces, there were more positive rewards than negative ones.

In this study, we achieved in situ patterning of on-chip gel actuators for high-throughput and versatile cell manipulations. Integrating actuators is essential for realizing high-throughput on-chip cell manipulation. Here, we successfully performed in situ patterning of on-chip gel actuators using direct laser writing (DLW) for high integration of the actuator. DLW enables precise placement of gel actuators on a microfluidic chip without manual alignment. As a result, many gel actuators can be integrated into the chip. Furthermore, we demonstrated the actuation of fabricated gel actuators as on-chip valves and success-fully achieved flow switching in a microchannel with high ON/OFF ratio.

A reciprocating linear actuation system utilizing electrorheological (ER) fluid clutches is developed. In this paper, we focus on achieving accurate position control of the proposed system. In order to realize real-time application, a simple feedforward neural network with a single hidden layer is examined to model the inverse problem of torque transmission. We then develop the position controller based on the computed torque control (CTC) framework, augmented with a proportional–derivative (PD) feedback loop. Experimental results confirm that the proposed control architecture delivers robust and accurate position control, even under disturbances caused by randomly changing motor torque inputs.

High-fidelity motion data are essential for imitation learning. This study reduces the control period of four-channel bilateral control from 2ms to 1ms by using a microcontroller and a direct RS-485 connection. Experiments on a single-degree-of-freedom (DoF) testbed show widened velocity and force bandwidth with higher gains under the faster control period. On an eight-DoF CRANE-X7 pair, four manipulation tasks executed at 1ms were up to 30% faster than at 2ms, with reduced variance observed in some tasks. These results suggest that shorter control periods enhance responsiveness, task efficiency, and demonstration data quality.