Koichi Ozaki’s research while affiliated with Utsunomiya University and other places

This study analyzes the recognition process of examining the texture of fine particles through tactile sensing using human fingers and aims at establishing the structure of texture recognition. Our previous study confirmed the occurrence of hysteresis in tactile sensation in the mid-teens. This study investigated the effect of the delay in sample presentation to the subject on the KANSEI evaluation of tactile sensation. The results showed that the KANSEI evaluation changed with the delay in the sample presentation.

We are conducting research to realize industrial operation of mobile robots. For position estimation, we adopted a method that uses two-dimensional information, keeping industrial use in mind and considering cost. Using this method, I participated in the Nakanoshima Extra Challenge 2024, completed the task, and completed the race. Through this experiment, we will discuss the characteristics of our position estimation method.

In SLAM, the distortion of maps due to discrepancies in revisited points is a significant problem. One solution is map correction through loop closure, which necessitates loop detection. LiDAR is often used for loop detection, but it can result in false detections of revisited points when similar features exist along the robot’s trajectory, making precise map creation challenging. In the case of large-scale SLAM, the likelihood of false detections rises, and memory costs escalate as well. To address these issues, we utilize WiFi signal strength information. WiFi signals offer advantages in terms of reduced false detection occurrences and lower memory costs, making them suitable for loop detection in large-scale SLAM. This paper applies loop detection using WiFi signal strength information to the map creation process of RBPF-SLAM, enabling the creation of accurate maps with relatively low memory costs.

In this manuscript, we propose a motion strategy for manipulating strings with unknown properties. Our approach iteratively refines its motion generation based on parameters estimated from observed string behavior, without the need for real-time feedback. This strategy has been shown effective in achieving several motion objectives using uniform strings of similar lengths. In this research, we improve upon this strategy by addressing the challenges posed by varying string lengths and non-uniform strings. For this, we utilize a non-uniform string model and address various string properties to demonstrate the feasibility of our proposed motion strategy. Experiments conducted with different string types and lengths (between 300 to 610mm), including some with non-uniform mass distributions, demonstrate our method’s effectiveness. Results show that our proposed method functions effectively with various kinds of strings, regardless of length and mass distribution, without requiring precise model parameters. Unique to this approach is its ability to adapt to various string characteristics through parameter estimation and motion generation, significantly reducing the need for real-world manipulation trials. Our findings illustrate the potential of our method for use in advanced robotic applications that require handling deformable objects.

Wheeled omnidirectional mobile robots have been developed for industrial and service applications. Conventional research on Omni wheel robots has mainly been directed toward point-symmetric wheel arrangements. However, more flexible asymmetric arrangements may be beneficial to prevent tipping over or to make the robot more compact. Asymmetry can also be the result of a motor/wheel failure in a robot with a redundant configuration; in this case, it may be possible to continue operations, but with an asymmetrical arrangement. For controlling such asymmetric arrangements, it is necessary to consider the moment of propulsive force generated by the wheels. Since it is difficult to measure the propulsive force accurately, in this work we model propulsive forces as being proportional to the ground speed of the wheels. Under this assumption, we estimated the robot’s behavior in an asymmetric wheel configuration by considering the balance of the velocity moment, which is the moment of the wheel’s ground speed. By verifying the robot’s behavior with various wheel configurations, we confirmed experimentally that the sum of the velocity moments affects the straightness of the robot and allows us to improve the design of asymmetric wheel arrangements and control during wheel failures.

Recent years, mobile robots are expected to provide security and transport in multi floor buildings such as condominiums and office buildings. To move between floors, robots need to use an elevator. From the viewpoint of labor shortage, robots need to be able to move to their destinations without human intervention. For this, robot itself needs to know the floor it is on. LiDAR is the most common sensor used in localization. However, with LiDAR, every floor could be similar data. So, need to use another sensor. Since WiFi signal strength varies with the floor of the building, WiFi receiving sensor is required for floor estimation. As a validation of the floor estimation we compare the likelihood of each floor in the entire area of the WiFi map using only WiFi information. As a result, we were able to estimate the floor where the robot is located in the building.