This paper describes the characteristics of joint mechanism that has an adjustable stiffness. In this joint, two types of silicone rubber cushions (SRC1, SRC2) are inserted between the motor and the output link. SRC1 is used to give a robot joint soft compliance with its elasticity. On the other hand, SRC2 controls joint stiffness by arranging the degree of its expansion. This expansion compresses SRC1 and makes SRC1 more rigid. We investigated the change in the joint stiffness according to the pressure in SRC2 through experiments.

View full abstract

Autonomous decentralized control is a key concept in our understanding of the mechanism underlying locomotion of animals, wherein the phasic pattern of motion and the excitation pattern of muscle tonus are well reconciled. In our recent work, we proposed a decentralized control methodology in which phasic control and tonic control are well coordinated. In this study, we develop a real serpentine robot to validate our proposed control methodology. Experimental results show that the robot exhibits highly adaptive behavior in response to environmental changes. The results obtained are expected to shed new light on the design of autonomous decentralized control systems.

View full abstract

We focused on the tip of legs that was the key to belly crawl flexible walking robot. This robot achieves walking based on flexible structure of the polyethylene plate and SMA. We considered the part of an insect's body that bears the wings and legs assumed to be insect's standard in the leg of the robot to improve the speed of the robot and applied an original flexible robot. We investigated the walking speed of developed robot by applying insect legs and influence on belly crawl flexible walking robot by angle of toe. Finally, we succeeded in the movement the twice earlier than an original robot.

View full abstract

This study proposes a new flexible mechanical system using hydraulic skeleton. With this system, flexible bags are used as hydraulic skeletons. In our former system, the output force is not high because of low durability of the bags which are made of soft urethane rubber. To increase the output force, this study reinforces the bags for durability enhancement. The bag is reinforced by polyester (C-type) or aramid (D-type). The rupture pressure is 0.80MPa in D-type and 0.44MPa in C-type whereas 0.06MPa without reinforcement type. Estimated maximum torque of D-type is 6.90Nm whereas maximum torque of the former type is 0.23Nm.

View full abstract

In this paper, we propose a shooting manipulation system with high reaching accuracy for manipulating a distant object for an instant. The proposed system mechanistically acts on a remote target by constraining an end-effector catapulted by an impulsive air flow through a flexible string connected with an elastic cantilever. Experimental results shows that the proposed shooting manipulation system has excellent basic performances such as high motion straightness of a reaching path, quickness in a turning motion, and high reaching accuracy in spite of its simplicity. Furthermore, we show that, by making full use of the basic performances of the proposed system, it is possible to realize quick capturing of a falling object 0.7[m] away with a high success rate of 92 percent.

View full abstract

In this paper, we propose a compact swimming robot powered by a new type of impulse force generator using closed elastica. One of the most distinguished features of the robot is to have capability to kicking water impulsively in spite of a compact body with low fluid resistance. The proposed robot is of 320[mm] in length and consists of a flat triangular body and an impulse force generator using closed elastica. The triangular body is of only 13[mm] in the maximum thickness and tapered for reducing its fluid resistance. The impulse force generator newly developed here, which is called 'spatially-deformable sharp type', is composed of a sharp elastic strip, two motors connected with an elastic bridge. When two motors rotate in the opposite direction respectively, the sharp part of the elastic strip passes thought its own closed loop and then a sharp elastic strip spatially deforms very rapidly due to snap-through buckling of the elastica. The proposed swimming robot can kick water strongly by fixing a fin to the sharp part of the elastica. We show an experimental result to illustrate that the swimming robot can swim 0.38[m] with one kick. We also mention possibility of realizing 2 [BL/s] (body length per second) swimming based on the robot.

View full abstract

In this paper, we propose a winged jump robot which has the mobility not only by jumping on the ground but also by gliding in the air from a height. The proposed robot consists of a trunk with a kicking leg, two wings, and a wing frame connecting the wings to the trunk. The trunk is based on the jump robot 'Robotic Kikker' developed by our laboratory which can jump at a distance of over 0.5[m]. The wing frame is designed not to transmit the trunk deformation during kicking motion to the wings. It is shown that, by selecting the size of the wings appropriately, the robot can reach at a distance of about 4[m] by jumping and gliding from a height of 2.6[m] at the maximum velocity of 3.9[m/s].

View full abstract

This paper discusses a nonprehensile dynamic manipulation of a deformable object, where the shape of a sheet-like rheological object is dynamically controlled by the combination of the inertial force and the frictional one generated by the plate's rapid motion. We first introduce a linear viscous model for approximating the object deformation characteristics, focusing on the final shape of the object. Assuming that the plate has two degrees of freedom: translational motion and rotational motion, we derive two sufficient conditions in order to deform the object: one to enlarge it and the other to contract it. Then, we show the cyclic plate motion leading to the continuous object's deformation, and finally, simulation and experimental results are shown in order to verify the proposed method.

View full abstract

In this paper, we explain a simple model and a deformation control of a linear flexible object with a high-speed robot arm. First, we propose the simple deformation model of a flexible rope using a high-speed robot motion. Then, we perform a theoretical analysis of the rope deformation model. And, we obtain a motion condition of the high-speed robot arm based on the theoretical analysis. Next, we show experimental results of the rope deformation control using the motion condition. Finally, we show the dynamic knotting of the flexible rope as an example of the rope deformation control.

View full abstract

This paper describes the optimization criteria of horizontal 2-joint Spring-Motor Coupling System (SMCS) focusing on the hand speed. In mechanical systems including motors and series springs, the angular velocity of an object driven by a motor via a spring is instantaneously higher than that of a system without any spring (Velocity Increasing Effect). We can obtain this effect on 2-joint serial link systems. We propose the maximization criteria of the hand speed of the horizontal 2-joint SMCS using steepest descent method and genetic algorithm. These criteria is useful for motions using explosive power such as a pitching motion.

View full abstract

In this study, we develop a vibration-driven arm system. Our arm system is comprised of a rigid part and a flexible part. The rigid part is connected to the flexible part loosely at node of vibration modes of the flexible part, and there is a brake device between them. The rigid part turns by vibration of the flexible part. The rotary direction and the rotary speed of the rigid part are controllable by the timing of brakes. Experimental results show the rigid part moves according to our theory. Our arm system are lightweight and safe because it don't need motors at its joint. We hope that this method allows the production of a simple and lightweight arm.

View full abstract

In this paper, we propose bilateral controls by using intervention impedance based on passivity about flexible master-slave system. We discuss about a control performance of the system based on the passivity.

View full abstract

In this paper, we present a novel eccentric paddle mechanism (ePaddle) for amphibious robots. With this mechanism, the robot will be able to work in terrestrial, aquatic and semi-aquatic environments and perform wheeled, legged and paddling locomotion gaits. The concept of this mechanism is briefly introduced. Then, the prototype design of the ePaddle mechanism is presented. Simulations on wheel-like gait and legged gait are performed, and the results verify the proposed concept of ePaddle with its prototype design.

View full abstract

To make the inspection robot pass in small pipes or vertical pipes, it is hard to equip several motors and mechanisms. Screw drive mechanisms have been focused because a minimum number of motor is only one. It can achieve downsizing, lightweight and energy-saving. Some existing robots can pass in the curved pipes with large curvature radius. However, it can not pass in the pipes with small curvature radius and mobility analysis and verification have not been conducted. In this paper, we consider the cross-sectional shape of a curved pipe that the in-pipe robot attaches. Necessary spring stiffness of elastic driving arms and motor torques are derived for the robot to pass in the curved pipe and the vertical pipe where the friction force changes.

View full abstract

It is important for a snake-like robot to have more available gaits for a higher adaptability. However, there are few snake-like robots to realize all of the common gaits under the current research framework. In this paper, a novel snake-like robot is proposed to resolve this question. At first, the snake-like gaits were analyzed to inspire the mechanism design. Based on the gait analysis, some non-snake-like mechanisms such as translating and rolling degree are integrated with the classical yaw joint. Next, the robot prototype was introduced and some basic experiments were performed to verify the expecting motion ability. This novel robot is expected to have a higher adaptability for its ability of multi gaits

View full abstract

Biologically inspired snake-like robot has been widely studied for its various motion patterns. To achieve autonomous adaptive locomotion, a sensor-driven snake-like robot has been developed. Such robot can achieve self-adaptive locomotion by coupling sensory information with the motion control system. The design of the mechanism and control system of this sensor-driven snake robot is presented in the paper. Based on the sensory information, the implementations of terrain-adaptive locomotion and collision-free locomotion are investigated respectively. The performance of the collision-free locomotion is verified by conducting an experiment on the robot in an environment with obstacles.

View full abstract

This paper presents a novel method for solving the motion stability problem of our omnidirectional robot, MY Wheel-II. For this robot, some inherent characteristics of the mechanism such as the different contact distances lead to the unstable and inaccurate motion of the robot. In this paper, a stable motion control has been achieved by improving both the wheel mechanism and the control strategy. An absolute encoder has been used to detect the position of wheels during the motion. According to different wheel positions, corresponding control commands are applied to the motors. Experimental results validate the effectiveness of the control scheme and the improved motion stability of the robot.

View full abstract

Human traffic line detection is required for many situations including security in office buildings and patient care in hospitals. However, realizing this function in real time is difficult because a large amount of computation is required for human traffic line detection that is tolerant of environmental changes including lighting conditions. In the present study, we have developed a system which detects human traffic line by using an analog silicon retina, which operates spatial processing in real time with low power dissipation. This system was successful in detecting human traffic line robustly.

View full abstract

We propose a method for personal identification based on hand-waving. The main contribution of this paper is how the people that wave a hand repeatedly can be identified using frequency features. We first construct a time series of hand low-resolution images. Next, frequency features are extracted from time series of hand low-resolution images applying FFT. Finally, our method computes the similarity of model data and test data using cone-restricted subspace method. In several experiments, we try to identify the person who waves a hand and show the accuracy of the proposed method.

View full abstract

This paper presents a novel method for gait-based person identification robust to changes in appearance. Gait is sensitive to appearance changes, such as variations of clothes and carrying conditions, so the correct classification rate is reduced in case target's appearance condition is different from that in the database. So we propose a new part-based person identification method, where the discrimination capability at each part is directly controlled based on gait features between gallery datasets and probe dataset. Experiments using a gait database CASIA show the effectiveness of the proposed method.

View full abstract

In this research, the purpose is making system of detecting human by using the thermography for Patrol Robots. Because they have different temperature between human and the environment, we can detect human by the thermography. We found the way to decide the range of temperature automatically to clip out the image of human from the thermography. And, We suggested the way to cut the elements beside human and proved the effect by experiment.

View full abstract

Retrieving a large collection of environment maps built by mapper robots is a key problem for mobile robot self-localization. This map retrieval problem is addressed from a novel perspective of a multi-scale Bag-Of-Features (BOF) approach in the paper. In general, multi-scale approach is advantageous in capturing both the global structure and the local details of a given map. On the other hand, BOF map retrieval is advantageous in its compact map representation as well as efficient map retrieval using an inverted file system. Combining the advantages of both approaches is the main contribution of this paper. Our approach is based on multi-cue BOF as well as BOF dimension reduction, and achieves efficiency and compactness of the map retrieval system. Retrieval performance is evaluated using a large collection of point feature maps.

View full abstract

In Japan, it is problem that the population decrease due to the low birthrate and the lack of labor power due to old people increase is developed. Recently, needs of the autonomous robot which can perform instead of human being have risen on production site. In this research, we decide on optimal camera position where camera can calculate the high-accuracy position. In this position, we verify the discrimination accuracy about sphere, triangular pyramid, regular hexahedron as objects.

View full abstract

There is a problem that the detecting and tracking method by using the image processing cannot the fast moving object. Many adaptive detecting and tracking method for fast moving object are proposed. However, these methods are not compared quantitatively because there are not the standardized evaluation methods. Moreover, the quantitative evaluations of these methods are not enough. In this paper, the quantitative evaluation methods are proposed and 2 methods are compared by the proposed methods. One of methods is the method using the HSV color space and updating the histogram. another is the MeanShift. The proposed methods consist of "Tracking Error," and "Limit Speed". The results of quantitatively Evaluation show that the proposed method is useful.

View full abstract

Time-of-Flight range finder gives a range value from the sensor to the target by measuring the time until the emitted laser pulse is reflected on the target's surface and returned to the sensor. The power of the reflected light is also measured as a side product of the range value. In this paper, we propose new inpainting technique for range images utilizing reflectivity. The range image inpainting technique based on Belief Propagation recovers a deteriorated range image using not only the adjacent range values but also the continuity of the reflectance image. We demonstrate that the proposed technique enables to repair deteriorated range images successfully.

View full abstract

In this paper, we propose a novel method for three-dimensional(3-D) object tracking based on integration of 3-D range and color data. To get 3-D range and color data simultaneously, our method utilizes the Microsoft Kinect sensor which is composed of depth and color cameras. 3-D range and color data are integrated using the estimated depth and color cameras intrinsic parameters and relative transform between the cameras. The target region can be tracked by processing depth pixels with color information. Through experiment in real environment, we confirm the validity of our method.

View full abstract

We have bulit a foveated camera employing two cameras, a perspective camera and a omnidirectional camera. The omnidirectional camera is used to build a peripheral view field, which has wide view angle. The perspective camera is used to build a central view field(fovea), which has high resolution. This camera arrangement can make easy to compute the direction to the target in the peripheral view field.

View full abstract

We developed a stereo vision system that can detect binocular disparity at a high frame rate. The system inspired by the visual signal processing in the retina and the visual cortex. Although perceiving depth is a requirement for robot vision, realizing this function in real time is difficult because it requires an enormous calculation. On the other hand, we living things can instantly perceive binocular disparity and depth with parallel and hierarchical structure in our visual nervous system. In the present study, we implemented a stereo vision inspired by neural processing on a hardware system with mixed analog and digital architecture. This system can perceive depth at 200fps.

View full abstract

In this paper, we report a method of observing dynamic flying objects such as a ball in a sport game stably and continuously. We have developed the Saccade Mirror, the high-speed optical gaze controller, and it has achieved an arbitrary gaze direction within ±30 [deg] in less than 3.5ms. However, even if a target tracking system is developed with it, some visible tracking errors still has been appeared mainly because of a computer's delay. We applied a real-time image processing to images obbtained by a target tracking system with the Saccade Mirror and resolved the tracking errors. Then, we could successfully observe an impact of a rubber ball by a racket.

View full abstract

The goal of this study is to develop a robot vision system which can recognize colors more precisely than human at high speed. To achieve this, we have developed a system which consists of a multi-spectral light source which can change its wavelength at high-speed and a high-speed monochrome camera. The system can recognize and track a target object at high speed. First, this system measures the spectral information of the target surface, and finds the features which are remarkable about spectral reflectance. Secondly, by changing the wavelength of illumination actively, the system tracks the target with the features of spectral reflectance at high speed. Experiment results, in which the system distinguishes a target from other objects which has very similar appearance in human eye, are shown.

View full abstract

We propose an algorithm to detect characters in images. The proposed algorithm consists of three phases: training, candidate area detection, and character detection. In training phase, a classifier to detect the character is generated from several haar-like feature-based weak classifiers by using Adaboost. In candidate area detection phase, one haar-like feature is used to define the area to search characters. And in character detection phase, the generated Adaboost-based classifier is used to detect the character in the defined area. We performed preliminary experiments to evaluate the proposed algorithm using 13 images. The experimental results show that some character on homogeneous background can be detected by this algorithm.

View full abstract

Sewer pipe exists close to our living environment and is one of the most important infrastructures, however, the inspection and maintenance of sewer pipes are not automated and is a risk for direct human operations. We have been working on automation of sewer pipe inspection using mobile robots. This paper describes a method of crack detection of sewage pipe using wavelet transform and categorizing the crack level.

View full abstract

Spark testing is normally conducted to determine and/or classify materials of steel stocks as a final step of quality controls in ironworks. However, there are not quantitative criteria and discrimination results are varied among inspectors even though spark testing can be conducted in short time and is nondestructive test. It also increases a sense of commitment and a burden of inspectors because he/she independently conducts a conclusive confirmation of quality controls which may affect company's confidence. Therefore, this paper proposes an automatic and a quantitative inspection method in spark testing by using image processing. In this method, carbon content of a steel stock is estimated based on amount of explosions and spark by extracting exploded spark which is one of indicators of carbon content.

View full abstract

This paper describes an evaluation of the L1-SAIF (L1 Submeter-class Augmentation with Integrity Function) augmentation broadcasted by Quasi-Zenith Satellite System (QZSS) in the urban environment for the vehicle positioning. In urban environment, it is expected that the augmentation of QZSS is quite effective because of its high elevation angle. Therefore, we carried out the static and moving evaluation test. In static evaluation test, we compared with the positioning error between the only GPS positioning and augmentation of QZSS positioning for evaluation of the positioning accuracy. In moving evaluation test, we compared with the visibility of QZSS and SBAS satellites for evaluation of availability. As the result of the evaluation test, it was confirmed that the augmentation of QZSS has better availability and effectiveness than the augmentation of SBAS.

View full abstract

This paper proposes accurate outlier detection method for GPS using 3D environment maps. Main topic of the paper is analysis of diffraction wave of GPS radio waves. GPS is major localization method for outdoor environments. In urban environments, localization results by GPS have large error caused by diffraction and reflection of radio waves. This paper describes the analysis of distance error of radio wave by diffraction, and outlier detection method for GPS with diffraction wave effects.

View full abstract

In this paper, we propose a learning method to solve the parameter determination problem of Divided Difference Filter (DDF) for accurate localization. DDF can achieve comparatively accurate localization than other Kalman Filter algorithms in poor GPS area. However, parameter determining process of DDF requires significant time and engineering cost. Furthermore, it is difficult to obtain optimal parameters for accurate localization by hand-tuning. DDF has three parameters which should be determined: covariance matrices of input and measurement noise and a Hyper-parameter. Our technique uses a discriminative learning method to determine these parameters. The proposal method absolves developers from the cumbersome process of parameter setting. This paper describes the efficiency of our technique through simulations and an experiment.

View full abstract

In this paper, we describe the easy and robustness self-position estimation technique against an unknown obstacle under the environment where many unknown obstacles exist. The free-space observation model is suggested as the technique of estimating a self-position in an environment with many unknown obstacles, such as a pedestrian. This has realized robust self-position estimation. However, the free-space observation model has a several problem that complicated likelihood evaluation, particle continues being spread and it cannot be converge when a immovable obstacle is able to observe either only right-and-left to a robot's direction of movement. In this research, we solve these problem with two likelihood evaluations that the based on the area of free space and based on the area of occupied space. We evaluate robustness and verify the validity of suggestion method by the simulation and the experiment under the real environment.

View full abstract

In this study, we aim to realize an autonomous mobile robot which works for a car. One of the essential functions to achieve this purpose is estimation of a relative position of the car to the robot. We propose a method of the relative position estimation of the robot by using the shape of the car measured with the laser range finder.

View full abstract

This paper describes a localization method with 3D map matching. There are various types of conventional localization methods, but relatively large number of conventional methods are weak on high spatial frequency shapes. In outdoor environments, there are many trees and bushes. These branches and leaves have complicated shapes, and the spatial frequency is high. Therefore localization methods adapted to high spatial frequency shapes are required for outdoor localization. We developed a new localization method with 3D map matching. In our method, the spatial frequency of point clouds is reduced, and broader directed edge points are extracted. Then ICDP algorithm matches directed edge points utilizing constraint of gravity direction. We evaluated localization errors on Tsukuba Challenge 2010. The maximum error was 0.675m, it was confirmed that our proposed method is accurate.

View full abstract

This paper presents the self-localization method using extended feature extraction method of the pole star algorithm which is used for satellites identify stars. The pole star algorithm extract feature from dot pattern, such as stars. This algorithm is high accuracy and computationally cheap. Therefore, pole star algorithm is suitable to be used in the dynamic environment, such as RoboCup Soccer. In this paper, extracting feature from line was enabled by building scan-line to detect lines into pole star algorithm. I conducted the evaluation experiment using a soccer robot in RoboCup Soccer. As a result, localization error was on average 138mm.

View full abstract

In this paper, we propose a 6DOF localization method by multiple resampling MCL for depth image sensors. Our purpose is development of a localization method by only using depth image sensors for mobile robots that cannot use motion sensors. Our proposed method executes sampling and re-sampling two or more times at the localization step. Proposed method enables to search a maximum likelihood position like hill-climbing search. Normal MCL executes sampling and resampling only once at the step. When there is no maximum likelihood position in samples, normal MCL is difficult to estimate a correct position. Our proposed method can estimate the position even in the difficult case by the searching algorithm. Experimental results demonstrate that the proposed method can robustly estimate 6DOF-positions without motion sensors.

View full abstract

This paper describes a forward velocity estimation method using visual information. Forward velocity estimated from ground surface images is not affected by wheel slip. Using the estimated velocity, estimation error of localization would be decreased. In order to estimate velocity accurately and stably, many feature points are extracted and tracked with ground surface images. Shifting of feature points in an image is transformed into movement distance of body by plane projection for calculating the forward velocity. Experimental results in a rough terrain environment like other planet's surface are presented for verifying our system.

View full abstract

This paper reports the fact that straight line patterns of tiled pavement are useful for localization of a mobile robot. The patterns are captured by a tilted camera which watches tiled pavement in front. View angle transformation is applied to the image of the tiled patterns in order to obtain right angles of the tile vertices. Applying Sobel edge detection filter to the transformed images yields edge orientations of the tiles. A peak of the edge orientation in the histogram of the orientations together with map of the absolute orientation of the tile edges tells heading yaw angle of the robot. This heading angle is incorporated into the odometry to correct the cumulative error by means of EKF. This process is continually iterated.

View full abstract

In this paper, a vision-based localization method for an outdoor mobile robot is proposed. Drastic illumination change is one of the significant issues on outdoor vision system. We employ High Dynamic Range (HDR) vision to improve the robustness of image acquisition to illumination conditions. Our proposed method uses 2D gridmaps containing salient line segments on the road detected in images. A particle filter is employed to fuse the gyro-assisted odometry and 2D grid map matching to estimate the position of the robot. We carried out localization experiments and long distance navigation experiments in Real World Robot Challenge (Tsukuba Challenge 2010). Our robot with the proposed method successfully navigated part of the course of Tsukuba Challenge 2010.

View full abstract

With recent progress in lifelong map-learning as well as in information sharing networks, compact representation of a large-size landmark database has become crucial. In this paper, we propose a compact binary code (e.g. 32bit code) landmark representation by employing the semantic hashing technique from web-scale image retrieval. We show how well such a binary representation achieves compactness of a landmark database while maintaining efficiency of the localization system. In our contribution, we investigate the cost-performance, the semantic gap, the saliency evaluation using the presented techniques as well as challenge to further reduce the resources (#bits) per landmark. Experiments using a high-speed car-like robot show promising results.

View full abstract

In this paper, we propose a mobile robot localization system using appearance based place recognition. Our system uses appearance based place recognition and dead reckoning with Wheel Odometry. We use the theory of the observation likelihood estimation introduced from Fast Appearance Based Mapping (FAB-MAP) for appearance based place recognition. Images collected by a robot produce Bag-of-Words representation and Chow Liu trees. This representation is based on quantized SURF detectors/descriptors. A certain place can be recognized when the collected image matches a reference image coming from the same place in scene database. In addition, problems of appearance based place recognition techniques are solved by using sensor fusion. Experimental results are presented in an outdoor environment. The proposed localization system can achieve the average position error within 25m without GPS in an outdoor environment.

View full abstract

In this paper, we evaluate methods to create Image Landmark Database. We define solution as two parts: Local Image Detection and Global Image Search, and design how process image stream in these parts. Experimental results shows that Local Image Detection worked well with combination of Optical Flow and some different feature detector, and Bag of Features was effective to image stream for Global Image Search.

View full abstract

Humanoid robots sensory signals typically suffer from noise. In the typical case of indoor environement, small obstacles like carpets, books, or wires can make the odomtry error degenerate which eventually results in severe inaccuracies during the localization process. In this paper, we describe a landmark based multi-robot localization architecture which handles robustly the self-localization problem of a team of small humanoid robots. The landmark detection under partial occlusion and affine transformations takes advantage of the real-time capabilities of a random ferns based vision system. The observations of a single robot and those of cooperating parteners are merged through a particle filter-based method. In our approach, the abslute localization of every single robot is achieved in a robot-centric way. The relative localization is kept by a remote machine. Since all the team data is interfaced via the remote machine, every robot can act independently from the rest of the robot network.

View full abstract

In this paper, we report an implementation of self-localization system for mobile robots running in outdoor environments where GPS data is sometimes not accurate enough. Our system is equipped with GPS, encoder and magnetic sensor. Data obtained by these sensors are integrated using Kalman filter. GPS observation is omitted when the robot is running through particular areas. These areas are determined based on standard deviation of the data observed in advance.

View full abstract

In order to make mobile robots move autonomously, localization problems have been widely researched. Therefore, several methods, such as an extended Kalman filter (EKF) method that is based on the Kalman filter (KF) and an unscented Kalman filter (UKF) that introduces an unscented transformation, have been proposed up to now. On the other hand, it is well known that smoothers can improve the accuracy of the estimation result of the KF. As one of typical smoothers, there is a two-filter form (TF) smoother. To apply it to nonlinear systems as well as the KF, an unscented TF smoother is being developed right now. In this paper, an unscented TF smoother is applied to a localization problem and the accuracy of position estimations is evaluated through simulations.

View full abstract