Taejin Kim

Relative pose estimation of underwater robot by fusing inertial sensors and optical image

Accurate pose estimation and control system is essential to achieve an underwater robot system for underwater explorations and task executions. This paper presents a method of pose estimation for underwater robots based on multi sensor fusion. The proposed method fuses inertial sensors and optical image. The inertial sensors are used to provide a basic estimation of the pose data. In addition, the optical image is used to correct acquired pose estimate of the inertial sensors. By using both inertial sensors and optical image, the proposed method can provide a reliable pose estimate for underwater robots. Experimental results verify the performance of the proposed method in a basin.

A simplified model based disturbance rejection control for highly accurate positioning of an underwater robot

In this paper, we proposed a disturbance rejection control method to improve the position keeping accuracy of an underwater robot. Not like a ground robot, the controlling of a floating body is a critical requirement to acquire high quality sensor data from, for examples, a gyro, an underwater optical camera, and an imaging sonar. The proposed controller consists of a model-based robust compensator called RIC(robust internal-loop compensator) in an internal-loop, and a conventional PID controller and a null motion controller in an external-loop, where the model we used in the RIC was designed by experimental data of a regulation motion. The proposed scheme was experimentally verified by an observation class ROV, iTurtle which has been developed by the Korea Research Institute of Ships and Ocean Engineering (KRISO) for underwater structure inspection. Results show much improvement in rejecting disturbances for position keeping.

Development of a ROV for visual inspection of harbor structures

Underwater harbor structures should be inspected regularly for safety problem. Conventional inspection methods which are performed by divers have several limitations including safety and health problem of divers. This paper presents development of a ROV and experimental results for visual inspection of harbor structures. The ROV is equipped with visual systems for inspection tasks, navigation system and thrusters. The ROV is operated manually to approach to target structures and acquires high quality images for inspection. A basin test and a sea trial were conducted to verify the performance of developed ROV.

A study on the effectiveness of MOOS-IvP for Unmanned Surface Vehicle

This paper explores the effectiveness of MOOS-IvP middleware for Unmanned Surface Vehicles(USV). As unmanned technology evolves, various functions are required to carry out some missions. Also, as the USVs have more sensors, the complexity of data sharing is increasing. A good solution to deal with this is middleware. The MOOS-IvP is a C++ based open source publish-subscribe middleware. It consists of two components. First one is MOOSDB which acts as a server. The other is MOOS application which acts as a client. Each applications corresponds to a sensor or an algorithm module. This paper presents a brief overview of MOOS-IvP, and examines the effectiveness with a case study for acquisition of sensors data. The case study results show that MOOS-IvP can be beneficial for developing a application software for USV.

Underwater robot exploration and identification using dual imaging sonar : Basin test

For the underwater search task, imaging sonar is very useful equipment to obtain the visual data of the underwater environment. Generally, the search range of imaging sonar depends on the frequency of the sound. High-frequency imaging sonar is suitable to search near area within 5m and low-frequency is able to search the area between 5m to 100m. In this paper, we presents experimental results of landmarks identification using dual imaging sonar for the autonomous underwater robot exploration. The underwater robot explores autonomously and estimates the robot location using inertial sensor data. Artificial landmarks are designed and detected by dual imaging sonar. If the landmark is recognized, it is mapped to the estimated robot map. To verify the suitability of the proposed method, we perform an experiment in basin using the underwater robot, ‘yShark’.

EKF SLAM using acoustic sources for autonomous underwater vehicle equipped with two hydrophones

Development of localization method is crucial for underwater vehicles to navigate in underwater environments. This paper presents a method of localization for underwater vehicles using acoustic signal. Artificial underwater acoustic sources are used as landmarks in underwater environment and underwater vehicle performs localization by acquiring acoustic signals transmitted from the sources. TDOA based method is used to acquire directional angle of acoustic sources with bayesian process. Then, EKF-SLAM is performed to estimate vehicle pose and locations of acoustic sources by fusing inertial sensors and directional angles of acoustic sources. Experimental results of a sea-trial verify the performance of the proposed method. The proposed method can provide a reliable estimation of the direction and location of the acoustic source, even under for a noisy acoustic signal. Experimental results verify the performance of the proposed method in a real sea environment.

Preliminary result on object shape reconstruction using an underwater forward-looking sonar

This paper proposes a practical object shape reconstruction method using an underwater forward-looking sonar. In order to achieve the object shape reconstruction, three practical methods are adopted. The forward-looking sonar has a vertical field of view, which affect the re-projection; thus, the vertical field of view is modified to a narrow angle to reduce the uncertainty in elevation angle. Simple noise filtering and range detection methods are designed and implemented clearly and concisely. A low pass filter is also adopted to estimate the probability of voxel occupancy. In order to evaluate the proposed method, object shape reconstruction was performed using the basin test, and the results of the experiment are reported.

Hierarchical topological modeling of marine environment

This paper presents a method of topological modeling for marine environment. The proposed method detects landmarks from underwater and above-water environment. The topological model is acquired by applying Delaunay triangulation to the detected landmarks. Nodes are defined as triangulated region of Delaunay triangulation and edges represent adjacency between two nodes. The proposed method constructs a hierarchical topological model considering detectable range of each sensor. It consists of two levels of topological maps. The first level map covers entire environment and each second level map represent details of each node of the first level map. The proposed method provides an efficient modeling method for marine environment. Experimental results are provided for a sea trial to verify the proposed environmental modeling method. The proposed method could be helpful for marine vehicles to perform given tasks.

Development of a multi-purpose unmanned surface vehicle and simulation comparison of path tracking methods

Recently, a number of unmanned surface vehicles (USVs) have been developed and researched for various fields such as military, environment and robotics. These vehicles are used as an application platform of diverse algorithms. This paper introduces a development of a multi-purpose USV “eVe-1”. Also, several path tracking methods are compared by using simulation. The hardware of “eVe-1” is designed as a catamaran. It consists of power system, control system, various sensors, and two thrusters. Functions of the software are collecting sensor data, processing algorithms for a mission, and transmitting information for remote control. Three tracking methods for navigation of USV are compared with each other: approximate linearization, input/output linearization, and nonlinear control. We carried out simulation of path tracking and experiment to demonstrate the performance of “eVe-1”.

Preliminary Result on Underwater 3-Dimensional Reconstruction Using Imaging Sonar

This paper proposes a practical 3D object reconstruction method using an underwater imaging sonar. In order to reconstruct the 3D object, three methods are utilized. Firstly, the vertical beam-angle of imaging sonar is modified to narrow angle to reduce an uncertainty of estimated 3D position. The wide vertical beam-angle makes the incorrect estimation result about the 3D position of the reconstructed object. Secondly, simple noise filtering and range detection methods are designed to extract a distance from the sonar image. Lastly, a low pass filter is adopted to estimate a probability of voxel occupancy. To demonstrate the proposed methods, 3D reconstruction for three sample objects was performed in a basin and results are explained.