This paper proposes a new way to build a tunnel inside the narrow and rugged space. Rescue robots are expected to be able to supply water and medicines to victims by using this tunnel. Key issues are how to steer the direction of the tunnel. The author proposed two methods to steer the direction of the tunnel. One is the method transforming the tunnel material by using the actuator installed in the head, and another is that the tunnel has the actuator function operated by the signal from the head unit The author made prototype and conducted the experiment of these methods, and showed the possibility of the direction steer.

View full abstract

A sarin gas incident occurred in the Tokyo subway system on March 20,1995. This fatal gas attack caused sicknesses to many passengers. Since the disaster broke out under the ground, the main job of rescue personnel was to transport victims on stretchers to a first-aid station on the ground. So, we research and development the rescue support stretcher robot for a heavy casualty disaster in underground areas or at stations can be handled in a quick, effective manner. The stretcher robot is expected to help firefighters achieve efficient rescue operations. In this paper, we describe the produced stretcher robot. And, we show mobility of stretcher robot.

View full abstract

We had developed FRIGO-D as a basic track robotic unit for research and development, especially for fire fighting and rescue robots. Some fire departments interested in our development, and requested that they needed to use FRIGO-D in an actual activity. However, FRIGO-D doesn't have enough performance for actual activities. Then, we developed FRIGO-M of heavy-duty FRIGO-D for actual fire fighting and rescue activities. After that, we had trial deployments of FRIGO-M for some fire departments. The fire departments checked FRIGO-M performance from the point of view that they use FRIGO-M in actual activities. We improved FRIGO-M's weak points based on their reports. Now, FRIGO-M is one of the useful track robotic units for actual fire fighting and rescue activities.

View full abstract

We aim at development of semi-autonomous control system for getting over unknown 3D rough terrain, and used Aladdin that is a 6DOF tracked vehicle with four flippers. In our previous researches, a sensor reflexive approach was used for the semi-autonomous control because a rubble environment shape suddenly changes. In this paper, the authors will propose a method of rollover avoidance for the tracked vehicle with nippers. The rollover avoidance composes of a prediction phase and a motion generation face. In the prediction phase, it is necessary to consider the crawler contact condition, four flipper s angles and the robot posture. The authors will use normalized stability margin from these sensor data for the judgment. In the motion generation, flippers will are controlled according to the flipper contact condition.

View full abstract

For space robots having manipulators many control methods have been proposed. Most of them use the inverse of Generalized Jacobian Matrix (GJM) which is a coefficient matrix between the end-effector's velocity and the joint velocity of the manipulator. We have proposed discrete time control methods using the transpose of GJM. The control methods use position and orientation errors between the desired and actual values of the end-tip of the manipulator. To obtain higher control performance we have proposed a digital trajectory tracking control method that has variable feedback gains depending on the desired linear and angular velocity. It is considered that joint velocity controllers are also used for space robot manipulators. In this paper, we propose a tracking control method for joint velocity controller. Experimental results using a simulator show the effectiveness of the control method.

View full abstract

One of the most important phases of satellite servicing operarions by a space robot is the "contact" phase. During the contact between the end-effector and grasping point, there is a risk that the target and the robot can be pushed away from each other by the contact force. Therefore, the method to control the end-effector by the impedance control so as not to deflect the target was suggested. However, all of real control systems have the control time delay. In case of the target capture operation, the time delay of the impedance control makes the impulse of the contact larger. In this paper, to velify the effect; of the time delay of the impedance control, experiments with a air-floating testbed and numerical simulations are carried out.

View full abstract

A hybrid motion simulator generally suffers from a problem in which the kinetic energy of the system increases during contact because of delay times. This paper presents a compensation method for that problem. A dead time model that represents various delay times of the system is presented, and how the kinetic energy of the system is increased by the dead time is clarified. On the basis of that, a method to compensate for the dead time is proposed. This compensation method realizes a variable coefficient of restitution in the hybrid motion simulation. After that, experiments of uniaxial collision are conducted to evaluate the proposed compensation.

View full abstract

Multi-Leg robot has some advantages about locomotive mechanisms at exploration on rough planet surface. This Paper shows basic control of multi-legs on Regular Octahedral Robot. Controls of robot legs have low dependency from their own posture.

View full abstract

This paper describes the lunar exploration rover model. In Japan the development of the rover to explore the surface of the moon is planned to get its detailed information. We developed the mechanical model of the lunar exploration rover with the crawlers and the suspensions, similar to the Rocker-bogie mechanism. The basic experimental results show its mobility on the uneven ground.

View full abstract

Lunar rovers are required to move on rough terrain such as in craters and rear cliffs where it is scientifically very important to explore. Some researchers are developed locomotion systems to move on surface like this. Surface to move by rovers are set some types. As types of surface to move, this paper sets soft soil. There are some problems to move on soft soil. We have to examine to move efficiency on soft soil. Therefore, we consider a mechanism between soil and wheel. The locomotion systems are able to develop using the knowledge get from this mechanism. This paper focuses on the active suspension systems which can produce effective driving force on soft soil. The effective contacting force are produced when the angle of links of the active suspension are changed to get effective driving force according to knowledge of terramechanics. This paper describes about the relation of terramechanics, and proposes active suspension system using the knowledeg of terramechanics.

View full abstract

Generally, wheels of a planetary rover slip easily and lose their traction on loose soil on the Moon or Mars. Hence, it is very important to keep traction for each wheel. Such interaction between a wheel and a soil is studied in the research area called "Terra-mechanics." In the Terra-mechanics, the traction force is calculated with soil parameters that are estimated under ideal configuration. However, real environment does not have ideal configuration constantly. Therefore, it is difficult to estimate an accurate traction force online. In this research, to estimate an accurate traction force, a wheel with built in force-sensors-array (BFSA) was developed. It acquires normal stress distribution online. By using such sensors, we performed measurement experiments in different slip ratio and different soils. Finally, one soil parameter which is difficult to estimate by way of conventional method was estimated by comparison between calculated traction forces (based on the BSFA wheel) and actual measured forces.

View full abstract

Methods of laying environment sensers such as lunerquake sensers under the regolith of the moon is wanted for the future luner landing mission. Because of low thermal conductivity of the regolith, it is quite thermally stable in the regolith. We propose novel mechanism for drilling into the regolith, which uses reaction torque of a wheel. We report the concept of the mechanism, the design of the prototype, the analysis model of the mecanism and the result of the test.

View full abstract

Recently, lunar exploration has been carried out by many countries. Two other ways are planned to investigate the moon. One is to clarify the whole image of noon by placing seismometers, The other is to have information by taking samples of regolith into the ground with excavating. The exploration robot is required that it can carry out missions e.g. excavating into the ground, taking a sample regolith by itself. We focused on peristaltic crawling of an earthworm whose locomotion mechanism can move while keeping a large area in contact with the ground. It is expected to be suitable for a lunar exploration robot. In this paper, we developed a peristaltic crawling robot that constitutes of parallel rink manipulator. Then we experimented basic movement of the robot.

View full abstract

To clarify the internal and global information of the Moon is very important in future lunar exploration. This paper describes a lunar subsurface exploration system to bury a seismometer in the lunar regolith by a burrowing robot. The authors have proposed a new screw drilling mechanism for a burrowing robot to achieve that purpose, in particular this paper describes the concept of expansion to the complete burrowing system applying the screw mechanism. The authors call the novel robot BSR (Burrowing Screw Robot). Also in this paper, the authors consider the requirements of the design concepts for developing the BSR and the experimental environments.

View full abstract

Many space missions are being undertaken to improve our understanding of the solar system and how to use space resources. Nowadays, since MER's success, exploration of planetary surfaces is being focused on. Rovers have the capability of reaching important scientific areas on planetary surfaces. Rovers traverse unknown environments in limited missions. Therefore, rovers are expected to have the ability to explore as widely as possible and to navigate themselves efficiently. The conventional navigation systems based on stereo vision or laser range finder are not wide enough to obtain distant area environment information. Since understanding wide environment makes rover navigation more efficient, detailed terrain maps are not necessarily required. This paper proposes an efficient image- based navigation by a single camera.

View full abstract

Investigations into small planetary bodies are becoming more attractive, and some missions to small bodies have already been sent. As such bodies have numerous geographical features, the next requirement is direct surface investigations by rovers. Localization of rovers is required to navigate them to specific points. However, conventional methods of localization are not adequate on small bodies because of the small body's irregular shape and small mass. We proposed an efficient method of localizing rovers on small bodies. It uses two-way range measurements between the rover and mother spacecraft. The measurements are conducted repeatedly. This method can be used over the whole surface of the investigating body, and it only needs one spacecraft. Numerical simulations demonstrated the effectiveness of the proposed method assuming an Itokawa-sized-asteroid, whose maximum diameter is 600[m]. In this paper, the sensitivity direction is analyzed, which indicates the preferable orbit of spacecraft and the sufficient number of measurement.

View full abstract

This paper describes about the construction of online simulation environment with which we can develop control algorithms for the track robots, mainly for the robot named "Kenaf." As a simulator, we use USARSim that can simulate 3 dimensional space. By employing the real time simulation of USARSim, we connect the real robot controller and the simulator directly and realize the online simulation with the real robot controller. We test some validations with the created simulation system and denote the availability of the system for the development of control algorithms.

View full abstract

In order to evaluate the rescue robots' performance in disaster areas by earthquake, it is necessary to analyze the charateristics of rubble with a large number of house structural elements. The paper deals with digital rubble models of wooden houses by collapsing process simulation using a physical engine. Different rubble models are constructed by changing the condition of the house structural parameters and their rubble characteristics are analyzed for accessibility of rubble's inside space by rescue robots.

View full abstract

The paper summarizes main results obtained in a five years research project on modeling and analysis of collapsed houses for supporting rescue activities. Study subjects on collapsed houses were categorized into five research fields: object modeling, forming process, analysis, human search methods and human rescue methods. As a key technology for connecting these fields, digital models of collapsed houses are studied from the aspects of measurement, construction, and analysis. Rubble forming processes are simulated with a large number of elements by using a physics engine and the spatial characteristics of the produced rubble models are analyzed in order to present the design requirements for developing rescue robots working in rubble.

View full abstract

IRS Soryu is multi trucks robot used for victim search in narrow area. It's difficult for an operator to control the robot using limited field of view image. Authors solve the problem using multi camera images. We developed the multi camera system and achieved the display of detailed and wide field of view image using multi camera image. However, much information is too difficult for human to understand intuitively. Therefore we need the simulator for changing some settings and verifying the method of display repeatedly. We developed the online simulation system which has following requirements; a. Online simulator of multi trucks robot, which considers its dynamics, b. Software frame works for getting multi camera images and transmitting them via socket communication in the simulator, c. Easiness to change some detail settings. Finally, we constructed the graphical user interface that enable human to observe detailed and wide field of view image in surroundings of IRS Soryu.

View full abstract

The PAS (Physical Agent System) is the remotely controlled autonomous robot. The PAS consists of an autonomousmobile robot subsystem (PAR: Physical Agent Robot), operation subsystem. In this paper, I propose smart vision platform and the hardware compensation method of camera image, using FPGA for PAS. The smart vision platform consist of FPGA, multiple cellular phone CMOS camera and embedded Linux.

View full abstract

Recentry, real-time flow measurement at high speed sampling rate is necessary for many fields. In this paper, we implement flow measurement algorithm on high-speed vision system, and reallize a real-time 1000 fps flow measurement system for megapixel image. We show its effectiveness by showing flow distributions measured at real-time.

View full abstract

Laser robot can weld more accurate than Arc welding. However, the accuracy of positional match is demanded. Then, the system the welding robot is equipped with the camera and slit beam irradiating to the welding object is assumed. And, the irradiated appearance is photographed. By giving geometrical analysis to the obtained beam image, laser robot is controlled. Here, as the preparation stage, about linear regression of beam for geometrical analysis is described.

View full abstract

In recent years, the robot which can operate in place of a person is needed in a factory. Therefore, in this research camera is way for a robot system. In this paper, a calibration method of image coordinates is presented. Heretofore, in conversion process from image coordinates to three dimensional location, stereo cameras must be parallel. However, in this time there are errors of camera setting, parallel setting of stereo camera is difficult. And so, we propose a calibration method of image coordinates for setting error of camera. At first, we will develop a calibration method for convergence angle of cameras. We need determination of baseline direction and convergence angle of two camera in order that we use this method. Then, we try method of determination of baseline direction and convergence angle of two cameras with landmark. Availability of this method comes out by experiments.

View full abstract

This paper described a person's head detection using the separability filter. Authors have succeeded in making a high sound area only in a local area at 50cm diameter level by using a speaker array. This is called a sound spot. In this research, this sound spot around a head is tried to form by using a vision. The person's head is detected by using an image obtained from a camera attached to the position higher than the person's head, such as a ceiling. The person's head is a circle when it is seen from over head. Therefore the head position is detected from an image by using the separability filter. And sound spot is formed to there.

View full abstract

A vision sensor with adaptive response was implemented. To maximize the amount of information expressed in each image, the sensor exploited logarithmic response to high intensity, and entropy based adaptation. The logarithmic response, which enhances the dynamic range of the sensor, was achieved by applying a logarithmically varying voltage to the reset terminals of the active pixel sensors. By adjusting the logarithmic voltage property, the sensor adapts to the environmental illumination so that the output image data distributes across the available voltage range. The appropriate logarithmic property was determined based on entropy maximization. The quality of the output image was maintained irrespective of illumination changes.

View full abstract

In tills paper, we propose a real-time abnormal motion detection system in high-speed repetitive operation using high-speed vision hardware. In addition, we introduce an algorithm to enable detecting high-speed abnormal motion in real-time. Experimental results show the effectiveness of the proposed system.

View full abstract

There is gait identification as the personal authentication technology. Gait identification to find an outsider and to recognize a suspicious figure has ever proposed. On the other hand, in order for the mobile robot to extend the place of activity, technology for person following is useful. So, in this paper, gait identification for person following is proposed.

View full abstract

A monocular optical distance sensing vision system was investigated that was based on an algorithm that utilizes motion parallax. The aim of this work is to build a new, robust, absolute distance sensing strategy suitable for unstructured and unprepared environments. In order to apply this strategy only one camera, which was a silicon retina sensor, was used to calculate a contrast enhanced input image together with a temporal change based on difference images on analog hardware. And Kalman filter was also used to decrease uncertainty from measurements and to do an optimal estimate. In this paper an experiment was performed where a static high contrast object was the primary target and the observer was accelerated with the help of gravity. The experimental data was compared with a Monte-Carlo simulation of the algorithm in order to help us build a noise distribution model of the system dynamics.

View full abstract

This paper describes the method that is accuracy improvement of the stereo vision system with available cameras. Generally, parallel stereo method is used in performing the stereo vision. However, improving the accuracy of 3-D reconstruction based on parallel stereo method, immense time and labor are required to conform internal and external parameters of right-and-left cameras. Especially, in not using the camera developed for robot visions, there is a limit in taking out high accuracy with the parallel stereo method. In case of differing internal and external parameters between cameras, This method enables 3-D reconstruction of high accuracy using epipolar geometry and the fundamental matrix.

View full abstract

First, We attempted to automate the process of analysis of American Football play that needs much time heretofore. We recognized the play from the video of American Football. Play recognition requires time-series positional data of players and lines. Trajectory of players is formed from them. We recognized the runnning pattern by comparing the trajectory to registerd data. And we recognized the defult position by using data of formation. Second, We attmpted to construct the system of play simulation. We simulated the play of defensive players. The defensive players in the simulator have AI that is the process of the thinking of the real defensive player of American Football. The offensive players in the simulator are made to run the set coruse. We replayed the result of simulation in the game map.

View full abstract

We propose a 2D microorganism tracking system using parallel level set method (LSM) and column parallel vision system (CPV). The proposed new energy function of LSM constrains the amount of light intensity inside detected object contours to control the number of detected objects. This algorithm is implemented in CPV system and computational time for each frame is 2 [ms], ap- proximately. Based on the single target detection and automated stage PD controlling, the single microorganism is tracked to be in the middle of visual field under a microscope. We demonstrated a tracking experiment for about 37 s which shows that a single paramecium can be kept tracking even if other paramecia appear in the visual field and contact with tracked paramecium.

View full abstract

This paper describes an efficient solution to a multi-camera view planning problem. We deal with the case where a relatively small number of active cameras track many persons for surveillance purposes. Planning camera view is therefore one of the important issues in realizing a competent surveillance system. In our previous paper, we formulated the planning problem as a combinatorial optimization problem. Since its search space is very large, we developed a multistart local search (MLS)-based planner with non-linear programming for initial solution generation. The method is efficient enough to be used on-line. We apply the method to the problem of selecting several focused area in the image sequence from a fixed camera, which simulates the multi-camera view planning problem. Considering the success ratio of CamShift-based tracking, our planning method is considered to be sufficiently effective.

View full abstract

This paper proposes a method to extract color information based on the feature area of a person's shape and how to use the information to track the target human from in real time. In this research, we perform the recognition through optimization and finding the solution using Genetic-Algorithm(GA), and the evaluation of the form of subjects using Model-based matching method. The parameters of shape and the color of the model are used to search in the gene of GA as a newly obtained information of that tracking human. Then the search model is continuousiy changed to the shape of the terget human to reflect and adapt to the individual feature in the dynamic image. It is shown that our method is able to extract the several features automatically by tracking experiments to human.

View full abstract

Visual extraction of an object that is the target of attention or manipulation, is an increasingly important issue in robot vision. In real world applications, a robot needs to extract an object designated by a human in a complicated environment. Most previous methods have a limited ability to extract desired objects from a cluttered scene. Moreover, for the purpose of human-robot interfaces, it is desirable to make it as easy as possible for the user to designate an object. In this paper, we propose a segmentation method that can correctly extract a target object in complicated real world visual situations. This method exploits color and distance information in an integrated way. The system requires only a simple input to designate the target object.

View full abstract

This paper proposes a 3D shape reconstruction method using a moving camera and a spherical mirror based on the Perspective Factorization Method. Factorization method is quite useful for 3D reconstruction from image sequences. However there is a strict condition that all feature points must be observed in all images. In our method, a spherical mirror is placed in front of a camera and reflected feature points are tracked even if the tracking feature points go out of view due to the camera motion. This paper presents a basic concept of the proposed method and some results of fundamental experiments.

View full abstract

We describe a silicon retina system for robust color detection against changes of illumination. A present model for color constancy is based on Land's retinex theory and opponent color theory. In this model, an input image is divided into red, green, and blue channels, and received by the image sensors with a logarithmic response. In each channel, the received image is filtered by a Laplacian-Gaussian spatial filter. And then, the opposite color responses, namely red-green and yellow-blue, are calculated. The specific color can be detected by specifying the area on the opponent color plane which is composed of two opposite color responses. We consider a hardware implementation of real time color vision system employing three wide dynamic range silicon retinas and FPGA.

View full abstract

To acquire images of a scene from multiple points of view simultaneously, the acquisition time of vision sensors should be synchronized. In this paper, an illumination-based synchronization technique derived from the phase-locked loop algorithm is proposed. Experimental results show that the sensor operation can be successfully locked to an intensity-modulated LED illumination signal as long as the feedback gain is carefully chosen.

View full abstract

We aim at development of a search robot in confined space. The authors are developing Kenaf that is 6D0F tracked vehicle with flippers for the purpose. One of main tasks for Kenaf is an accurate 3D map construction in the confined space. In this paper, the authors will explain our approach for the 3D map construction, and will show the experimental result in Sendai Subway, Dec. 2007. For the 3D map construction in Sendai Subway, we needed to satisfy followings three requirements: 1. small size crawler robot that can get over steps, 2. small-size 3D scanner that can measure uniform and wide area's shape, 3. ad-hoc networks that can cover wide area. We satisfied these three requirements using Kenaf, TK Scanner and Wireless LAN respectively. Contribution of this paper is that we used our proposed method in the real environment and built a 3D map of Sendai Subway.

View full abstract

The robot which can be active outdoors is demanded with development of robot technology. Therefore we developed "Beetle-One" of outdoor mobile robot in the environment that is a sidewalk and a town area.An environmental map is necessary for autonomous run evasion in a person and an obstacle and find self-position identification.as a generation system of an environmental map ,aline and align a filler the data which acquired during a run one by one. However, measurement points are different, and cause an error of alining data.In this paper, indicate a system of Beetle-One ,and using global map generation method by iterating overlap-part estimation discussed and evaluated by some experimental results.

View full abstract

We were planning a technical challenge named "Tsukuba Challenge" last November in Tsukuba city. It is a part of the Real World Robot Challenge which is organized by robotic researchers to realize actual autonomous intelligent robots in our daily life. In the Tsukuba Challenge, robots are required to run autonomously and safely 1 km on a city promenade where people are daily walking. It is perfect Real World not specified places for robots such as laboratories, robot fields, and etc. This is the first trial with the actual world in Japan. In this paper the "Tsukuba Challenge" is briefly in produced with historical background of "Micro Mouse Contest"

View full abstract

Humans build a cognitive map by extracting geometry and semantic information from the real world. This map plays a significant role in navigation and communication. In this paper, we propose an algorithm for mobile robots for building cognitive maps autonomously. The robot observes the environment with a stereo camera to detect regions that are likely to include text information. If the robot finds such region, the zoom camera zooms in that region to estimate the probability of text existence more precisely. If the probability is high, the robot approaches there and extracts the text information by using the proposed algorithm. The robot builds a cognitive map by storing both text and geometry information. We conducted several experiments to show the validity of our proposed algorithm.

View full abstract

For the future cohabitation of humans and robots, we are making research on home use robots, where positions of robot, human and environment must be precisely recognized. We used omnidirectional mobile robot with laser range sensor for this purpose, and made experiments on map building and human following. First, we adopted scan matching algorithm for the localization of mobile robot in the stable environment, and confirmed the performance of the map building by experiments. Next, we applied this algorithm to the mixed environment of human and robot, and succeeded to extract human motion and robot motion along with the map building at the same time. We implemented the motion extraction algorithm for the human following application.

View full abstract

In this research, we achieved the real time LRF-SLAM system based on ICP scan matching and GraphSLAM by Lu and Milios. As the GraphSLAM algorithm which requires long computation time is calculated in sub thread, our system can localize in real time. In addition, because our scan matching method can detect occlusion areas and moving obstacles, it is robust on dynamic environments.

View full abstract

It is necessary for a robot to have environmental information in order to move autonomously from the current position to a given destination. Many research studies have been conducted on the navigation method for the autonomous mobile robot and introduced precise and accurate robot self-localization methods. However, we believe that in some situations, it is not necessary for a robot to precisely and accurately identify and measure its own position. In this paper, we proposed a robust robot self-localization method where robot will identify its own position based on visual features in environment. The experimental results demonstrate the effectiveness of our proposed method.

View full abstract

In localization of indoor mobile robots, dead-reckoning method is generally used. However, dead-reckoning method is known that it is unreliable because of uneven floors or wheel slippage. It is therefore necessary to augment the dead-reckoning method by measuring robot's position relative to known objects in the environment. In the present study, setting up landmarks in the environment is known as a simple method. However, setting up landmarks is hard work and requires expenses. In this paper, we proposed a method of using DC magnetic field occurred in the environment to localize robot's position. In this research, we use a magnetic sensor to read DC magnetic field in the environment, and use it to build a magnetic map. By matching magnetic map with magnetic sensor's readings, the robot is possible to accurately localize.

View full abstract

The spiral radius of a trajectory is most useful information to navigate a mobile robot in traveling on a spiral slope. The measurement method of this spiral radius has not been revealed yet. This paper describes the measurement method by some calculation of robot inclination sensor data. The spiral radius is related to the maximum inclination angle of a spiral slope and the maximum inclination angle can be detected by the combination of roll angle and pitch angle data. The experiment of the trajectory control on a spiral slope using the detected spiral radius is achieved successfully.

View full abstract

This paper presents a estimation for localization in outdoor environment. In outdoor environment, EKF(Extended Kalman Filter) generically reduces presumed accuracy for a remarkable change of sensor accuracy. Then, we built the system which maintains presumed accuracy by changing the parameter of EKF from the reliabikity of sensor data dynamically. Our system was tested using autonomous mobile robot "INFANT"(Intelligence Foundations for Advanced Navigation Technology) developed at our laboratory in an outdoor environment surrounded building. The sensors used in this test are GPS(Global Positioning System), a magnetic direction sensor, a rate gyroscope, and a pulse encoder. In this test, it drew the presumed result more reliable than the usual EKF.

View full abstract

This paper reports and outdoor localization method using a multi-sensor approach. Sensors were mounted on a differential drive Yamabico mobile robot. Experiments were done in university campus paths which are surrounded by trees. The localization framework is composed of two wheel rotary encoders, an Inertial Measurement Unit and a GPS. Localization approach using an Extended Kalman Filter (EKF) for data fusion is explained as well as data outlier rejection approach using a Chi-square test. Finally, experimental results validate the effectiveness of the proposed method.

View full abstract

In this paper, we propose a fast and reliable scan matching method so that it can be used in the cluttered environment. To speed up the processing time, our method utilize coarse-to-fine strategy. First, our method employs histgram matching approach to estimate displacements of position and orientation roughly. Second, to estimate them accurately, our method employ constrained least square method based on the formulation using poin-to-line metric. Such metric is also effective to the cluttered environment. The experimental results show the usefulness of our method.

View full abstract