Service robots need to be able to avoid pedestrians safely and reliably in environments where they coexist with pedestrians. However, it is difficult for robots to detect and avoid pedestrians from a safe distance because pedestrians at the end of a corridor or around a corner are in blind spots that the robot cannot detect with its external sensors. Therefore, this research proposes a robot navigation method that detects pedestrians in the robot's blind spot by receiving Bluetooth low energy (BLE) Beacon signals emitted by Bluetooth devices such as smartphones and wireless earphones that pedestrians generally have, and safely avoids them. By using the proposed method, our mobile robot can detect pedestrians that suddenly appear from a corner of a corridor earlier than when using a laser rangefinder, and avoid them safely in our preliminary experiment.

Japan is facing rapid population aging, and age-related decline in mobility directly impacts activities of daily living (ADL). Sit-to-stand (STS) motion is a fundamental movement, and difficulty in performing it can reduce activity levels, accelerate muscle weakness, and increase care needs. Conventional assistive devices mainly provide mechanical support and rarely encourage voluntary muscle use, which may result in further functional decline. This study proposes STS support robot designed to assist elderly individuals with slight care needs while promoting the use of their residual muscle strength. The robot generates assistance trajectories based on a stability margin, defined as the set of postures that enable safe STS motion using the user’s remaining capabilities. By associating patterns of muscle weakness with corresponding stability margin shapes, the proposed approach allows personalized assistance without the need for complex musculoskeletal simulations. Experiments with healthy subjects confirmed that the system maintains balance and encourages voluntary engagement of residual strength. This approach is expected to contribute to the prevention of further functional decline in elderly users.

This study aims to develop a system that visualizes pedestrians located in non-line-of-sight (NLOS) regions—such as intersections and shadows cast by buildings—and presents this information to users through VR headsets, thereby reducing collision risk and supporting safe mobility. As the first phase of this research, an image processing system was developed to analyze camera footage and estimate the real-time positions, postures, and movement directions of both pedestrians and users. Specifically, a method was constructed that employs bird’s-eye view transformation for accurate pedestrian localization and utilizes skeleton data derived from pose estimation to recognize postures and movement directions. Evaluation experiments demonstrated high accuracy for basic movements and confirmed the potential for future three-dimensional reconstruction.

This study investigates a navigation method for service robots based not on conventional geometric information, but on natural language–based path generation. First, route planning is performed using OpenStreetMap data, and human-readable instructions such as "turn right at the intersection" are generated from the resulting path. The proposed method is then evaluated in terms of labor and processing time by comparing it with existing GPS- and SLAM-based approaches.

This paper proposes a lower limb rehabilitation robot for normal walking in post-stroke hemiplegic patients. Ergometers are commonly used for lower limb rehabilitation, but their use has depended on the experience of physiotherapists. Therefore, this paper proposes a new rehabilitation robot based on an ergometer that leads the patient to voluntary muscle activity and learns how to use the muscles required for normal walking. The paper first compares how pedestrians walk between a normal person and a hemiplegic patient using muscle synergy analysis to identify the lower limb usage required for normal gait. The paper then proposes a method to approach the use of the lower limb required for normal gait by changing the load depending on the position of the pedals when a hemiplegic patient uses the ergometer. This will enable stroke hemiplegic patients to train for normal gait by pedaling the ergometer. The above proposed method has been implemented in our prototype robot and the effectiveness of our proposed method has been confirmed in experiments with post-stroke hemiplegic patients as subjects.

This study proposes the hands-free crutch featuring the load-sensitive locking joint. Traditional crutches often hinder the user's natural gait and pose risks of nerve paralysis due to pressure on the hands and armpits. In addressing these issues, we propose the hands-free crutch equipped with passive joints to facilitate a more natural gait. This paper outlines the development of the load-sensitive locking joint, which enhances safety by complementing the passive joint. Furthermore, we assess the performance of the proposed crutch by measuring joint and knee angles and walking symmetry by the symmetry index.

Overhead work in many industries causes a heavy burden on the neck because the head is always facing upward, and fatigue tends to accumulate in the neck and shoulders. In order to solve this problem, this study proposes the development of a head looking up posture support device to reduce the burden on the neck during upper overhead work. In this paper, we propose the device together with experiments in which the designed device is attached to a subject and electromyograms are measured. The experimental results show that the device can reduce the burden on the neck.

The aim of this paper is to realize a robot that supports standing motion for elderly people with low care needs, using the person's residual physical strength to prevent further decline in their physical strength. To realize this objective, the two key ideas proposed in this paper are as follows. The first idea is to identify the multiple basic movements that make up the standing motion of the elderly through a muscle synergy analysis of the standing motion of the elderly. This makes it possible to design standing assistance procedures according to the physical condition of the elderly person, as the basic motions can be matched to the stability margin range, which varies from one elderly person to another. The second idea proposes standing assistance procedures that allow the voluntary standing motion of the elderly to demonstrate their remaining physical abilities in standing motions. Even if an elderly person performs a standing assistance procedure that differs from the designed standing assistance procedure, our standing assistance robot allows the posture of the elderly person if it is within the stability margin of the individual elderly person, and only gives force assistance to the elderly person if the posture is not within the stability margin. This leads to safety and voluntary movement during standing motion. The above proposed method was implemented in our standing assistance robot prototype, and through experiments with subjects, we confirmed that it brought out the subject's remaining physical strength in standing assistance procedures.

The purpose of this study is to develop an archery practice system that does not require a long range. The core of the system is the estimation of the arrow's trajectory. To estimate the arrow's trajectory, we need to know the position of the arrow's passage in a plane perpendicular to the arrow's direction of travel at two points in the trajectory, and the time required for the passage between these two points. In this paper, we propose a method for estimating the position of the arrow's passage on a plane perpendicular to the arrow's direction of travel. The proposed system consists of a pair of object detection sensors, consisting of infrared LEDs and phototransistors, arranged in array on the opposite sides of a rectangle flame with length 600mm and width 150mm. This one-dimensional array of optical sensors can estimate two-dimensional position information. This paper reports the results of basic experiments.

This paper quantitatively analyses the effects of training with an ergometer by analyzing muscle synergies of muscle movements that appear in patients during gait function recovery training using an ergometer. By setting the load of the ergometer appropriately, the muscle synergies that appear when healthy participant train with the ergometer have a close similarity to the muscle synergies that appear during walking, suggesting that training with the ergometer is useful for the recovery of walking function. Furthermore, when stroke patients trained with the ergometer, the paralyzed leg did not show muscle synergies that were common to those during walking. This was therefore found to be due to the structure of the ergometer, which is operated using both legs, and the healthy leg compensating for the movement on the side of the paralyzed leg.

The overall goal of this research project is a study using RALL (Robot Assisted Language Learning) for Japanese students to improve pronunciation and prosody skills, and to provide more evidence of the effect of embodiment on RALL in English pronunciation and prosody. The subjects are divided into two groups: ten participants who use a robot (RALL) and ten participants who use a screen-based e-learning via computer (non-RALL). The robot had been used for the experiment was NAO from Aldebaran. The experiment took around 3 weeks for one participant to complete all the lessons including three pronunciation lessons and two prosody lessons. In order to assess the experiment, independent t-tests were performed regarding comparison of RALL and non-RALL scores difference. Dependent t-tests were performed to assess the performance of each group. The results shows that there was no significant effect of embodiment between RALL and non-RALL group. However, RALL system could improve Japanese students` English pronunciation, prosody, and overall pronunciation and prosody aspect better than non-RALL system.

The steerability of shopping cart varies with loading weight or position, affecting comfort and safety. The proposed active steering caster offers constant steerability of a cart. This caster consists of an electromagnetic clutch and a motor. The steering shaft is driven by the motor, and the transmission of that drive can be switched by the clutch. These casters are mounted at two diagonal points on the cart. By controlling the steering shaft of these casters according to the user's pushing force on the cart, constant steerability can be provided without being affected by the loading weight or position. In this paper, we propose the active steering caster with passive function and its arrangement on the cart with mixing the conventional casters.

This paper proposes a new estimation method for the lower limb muscle strength of the elderly that can be easily performed in nursing care facilities. Furthermore, based on the estimated lower limb muscle strength information, this paper proposes a human model that simulates the physical conditions of the elderly and, using this model, proposes a new design method for standing assistance strategy adapted to the physical conditions of the elderly. The main advantage of the proposed method is that it can estimate and model leg muscle strength, which depends on the individual elderly person, in a simple way that can be carried out by staff in nursing care facilities who are not engineering specialists, without the use of specialized measuring equipment such as electromyography.

In recent years, development of robots that can run independently in the same space as humans has been progressing. Most of the current navigation methods for autonomous robots use geometric information. While this method is easy to implement, it has some drawbacks, such as the need for accurate self-positioning and vulnerability to errors in sensor information. Humans do not have an exact self-position, but rather, they recognize an approximate location based on surrounding information and envision the path to the goal to reach the destination. Therefore, this study examines a navigation method that plans a route based on an abstract map, such as a hand-drawn map, and then travels independently to the destination.

In this paper, we propose an autonomous mobile robot navigation system that considers pedestrian flow in a crosswalk. In recent years, autonomous mobile robots have been required to be introduced into the real world. Such a robot is required to move efficiently, but more than that, it is required to avoid pedestrians and obstacles safely. On a normal sidewalk, if there is a risk of a collision with a pedestrian, the mobile robot can wait until this danger is gone. However, at crosswalks, there is a risk of collision with pedestrians due to multiple pedestrians coming and going, furthermore there is a time constraint of having to cross while the signal is green. Therefore, conventional mobile robot navigation methods cannot be used in this situation. On the other hand, the pedestrians move in the direction of each other in response to each other's movements, and this walking way enables a large number of people to pass through the crosswalk in a very short period of time, when the signal is green. Therefore, this study analyzes the movement of pedestrians passing through the crosswalk in a group, and investigates a mechanism for the generation of pedestrian flow on the crosswalk. Furthermore, we propose a mobile robot navigation method that can both avoid the risk of collisions with pedestrians and pass through crosswalks while the signal is green, using an algorithm to guide a mobile robot along the pedestrian flow that we investigated.

This study develops the robotic IV pole and proposed forward moving scheme aiming to reduce the risk of falling. The risk of falling caused by pushing conventional IV pole while walking can be reduced by having the IV pole autonomously move side by side with user (user does not touch IV pole.). The Robotic IV pole is composed of omni-wheel, Light Detection And Ranging (LiDAR), Raspberry Pi, and Arduino Due. The proposed method that user's center of gravity can be obtained by LiDAR being attached to the lower center of the robot base of the IV pole, detects both feet of the user, and each center of gravity positions of both feet is estimated. Then, the user's center of gravity is estimated by using the estimated center of gravity positions of both feet. By controlling the robotic IV pole's motion so as that the estimated user's center of gravity and the position of IV pole be the same in forward direction, the robotic IV pole can autonomously move with the user. By conducting experiment, it is confirmed whether the proposed algorithm can automatically move side by side with user, and how range of walking speed is acceptable. As a result, the IV pole was able to automatically moves with the user at walking speed of about 0.3 m/s to 0.8 m/s.

In this paper, we describe the integrated system for ensuring the safety of forestry workers. It is difficult to accurately acquire GPS information in forests. Therefore, it is necessary to consider a position estimation method that does not use GPS. First, the accuracy of VisualSLAM, IMU, and Beacon alone was verified and evaluated by experiments. We also examined the communication method and self-position estimation system for the integrated system. The results show that the self-positioning of Visual SLAM and IMU requires a large interval for correction, which results in a large error. Therefore, we found that the larger the measurement interval, the larger the error of any sensor system. The future tasks are to improve the accuracy of sensor systems, to optimize the measurement interval, and to integrate the systems.

This paper, we describe a method to detect glass existing in the real environment. One of the characteristics of glass is the semi-mirror surface. Focusing on its characteristics, we investigated a method for detecting glass from itself and its own reflection image. From the results, we were able to detect the glass by the proposed method.

The following topics are dealt with: learning (artificial intelligence); user interfaces; human computer interaction; artificial intelligence; medical signal processing; service robots; image classification; man-machine systems; mobile computing; and feature extraction.

This research discusses a dialogue system for human-robot communication. In order to develop a dialogue system, we conducted prototypes, experiments, and evaluations for each of the following three elements, “voice recognition”, “sentence comprehension”, and “response sentence generation” that are performed when a human talks with another person. Each element could be activated even though it was in the prototype stage.

This paper proposes a novel assistive walker robot, which considers its user's gait. Previous researchers have been developed many assistive walker, and many of them have discussed on obstacles/step avoidance function, path following function and gravity negating gravitational force function. However, by opinions of nursing specialists, the walker is first required to realize safe walking function especially, which prevents falling down of its user. Therefore, it is important to maintain the body balance of its user and considering with the characteristics of its user, the walker should coordinate it according to the walking way of its user. For realizing this, our proposed walker measures the right and left foot gait of its user and estimates its user's body balance by calculating his/her position of center of gravity (COG). Proposed system does not require sensors which are installed on its user's body for practical usage. Using estimated body balance, our proposed walker controls assistance force vector for maintaining body balance stably. A prototype assistive walker, based on the proposed idea is fabricated to help patients walking safely according to their walking gait.

This paper proposes a novel sitting assistance robot, which considers the posture tolerance of its user. The standing and sitting motion are different essentially because the standing is lifting motion against gravitational force and sitting is posture coordination to sitting position according to the gravity. Therefore, the robot should lead the patient’s posture within a stable range during sitting and the required performance is different from standing assistance. However, in previous studies, conventional assistive robots used the sitting motion which is “reverse” motion of standing. Furthermore, these robots helped patients by using a fixed motion reference pathway in spite of their original intention, and as the results, these robots failed to assist by confliction between their intended motion and reference path. Therefore, we propose a novel sitting assistance robot, which allows its user to move their body within a prescribed degree of posture tolerance during the process of moving from a standing to a sitting position. Our key findings cover two fundamental research topics. One is the investigation into posture tolerance during a sitting motion. The other topic is a novel assistance control algorithm that considers the investigated posture tolerance by combining position control and force control. A prototype assistive robot, based on the proposed idea was fabricated to help patients sitting down safely according to their original intention.

This paper proposes a novel standing assistance scheme that guides a subject to adopt a suitable posture using intended movements and voice messages. In much previous research, an assistive robot requires the patient to understand the standing reference beforehand and generally, the subject learns the standing motion used by the robot. However, for practical use, a patient should be able to use an assistive robot without special prior knowledge. Nursing specialists also guide their patients with an intended movement and a voice message. For example, when they incline a patient's upper body before lifting it up, they notify the patient of a motion start with simpler words and guide the patient's body with a special moving pattern. Therefore, in this paper, we develop a standing assistance scheme that guides the patient to stand in the intended way based on skills of these nursing specialists. The effectiveness of the proposed method is verified by experiments with our prototype.

In this research, in order to clarify robot design elements with high affinity with people, we developed an evaluation system that enables users to evaluate impressions from a highly flexible perspective from augmented reality. Human investigated impression evaluation on robot and identified 5 factors which are the evaluation index.

The purpose of this research is to develop the agricultural support system aimed at light labor and efficiency when doing agricultural work. The support system attempts to achieve the goal by proposing appropriate actions to humans from the state of agricultural land and the growing situation of the crops. To support agricultural work, this system transmits weather data, environmental data of agricultural land to the web server by sensors and scraping. This data is received with the created application. Following the value of the received data, the support system proposes the appropriate actions for the human. The proposed action decided from the four factors of temperature, humidity, illuminance, and soil moisture content, respectively. The experiment is the subject grow crops. In addition, confirm the significant difference between using the support system, not using it. After the experiment, analyze the questionnaire result by T-test to confirm the significant difference. From the results of the experiment, a significant difference in experiments confirmed by the t-test. Therefore, humans grow crops using proposed actions from support systems, It confirmed that light labor and efficiency realized. (183 Words).

In this paper, we considered the localization methods which recognizes landscape. In order to realize localization technique by recognizing landscape like a human, we recognize the landscape from objects arrangement in view. In this paper, we consider landscape recognition by SVM and segmentation method by deep learning. As results of object recognition by SVM and segmentation, the recognition rate was higher in the method by segmentation. In addition, from the experiment of localization by landscape recognition at multiple locations, we clarified the problem when applying this method to localization.

In recent years, service robots have advanced toward having to the ability to move autonomously in an environment in which they co-exist with humans. Thus, the robots must engage in obstacle avoidance to arrive at their goal point. In our previous research, we developed the artificial potential based method that the robot to plans an effective path simulating pedestrian characteristics by gently changing its moving distance. The repulsive potential changes depend on the pedestrian's characteristics. Using our idea, the robot can plan a path without including unnecessary potential changes, according to the pedestrian's route estimation range. Thus, our idea realizes both an effective path and pedestrian avoidance. In this method, the characteristics of the pedestrian are constant. However, pedestrian's walking characteristics are different depending on the environment that pedestrian's walking direction is various in free space model like shopping mall, in contrast, they are determined along the route in corridor model like an office. Therefore, the pedestrian's characteristic model should be changed depend on environment. Because the number of the avoiding-path case is unlimited, we propose the pedestrian potential model is applicable for any environments in this study. In the proposed method, the robot determines the pedestrian's position and plans the avoiding-path, which is reflected in the pedestrian's repulsive potential spread. The value of spread is calculated and changes depending on the pedestrian's position in real-time. We confirmed the proposed method is efficient by comparing experiment with the previous method, where the spread potential constant. The proposed method's effectiveness was confirmed through experiment with the prototype robot.

This paper proposes a novel driving-assistance system for manual wheelchairs with consideration of both uphill and downhill conditions. On an inclined road, there is a high risk of a wheelchair moving in a direction that the user does not intend. In our previous works, the user has driven our assistive wheelchair in the usual manner. Our proposed system estimates its user's intentions and passively works to complement their intentional force by negating the wheel traction that is generated by the road's inclination using only the servo brakes on each wheel. Nevertheless, in some cases, our system fails to assist the driving motion of its user because the user drives the wheelchair in several ways that depend upon the environmental condition, for example, during uphill or downhill driving. The required assistance is not constant according to the situation, and it is difficult to assist with one wheel-control algorithm. Therefore, in this study, we first investigate the required assistance condition according to the driving situation by conducting a preliminary experiment with wheelchair users. Considering the results of this investigation, we then propose a novel user interface that intuitively shows the system information and a wheel-control algorithm that selects a suitable wheel controller according to the driving situation.

This paper proposes a method for detecting and tracking pedestrians, which is an important function for preventing the collision of autonomously moving mobile robots with a pedestrian. To observe the pedestrian, we use two types of laser range scanners: one scans a 2D plane (2D scanner) and the other scans a 3D space, such as a 3D scanner. We scan 3D space with a 2D scanner fixed on the turning mechanism, instead of a 3D scanner, because 3D scanners are more expensive than 2D scanners. The 3D space is divided into voxels to determine whether a pedestrian exists. This detection system is robust to occlusion because turning 3D scanner can detect a pedestrian based on the shape of the data points. However, the scanning span of the turning 3D scanner depends on the period of the turning mechanism; thus, we combine the detected pedestrian's position with the data scanned by the normal 2D scanner because it is possible for a 2D scanner to track objects moving in real time according to the positional transition of the data points. To investigate the efficiency of the proposed method for pedestrian detection, which is robust to occlusion and exhibits real-time tracking using reasonable scanning devices, we experiment with our prototype.

This paper proposes a novel driving-assistance system for manual wheelchairs with consideration of both uphill and downhill conditions. On an inclined road, there is a high risk of a wheelchair moving in a direction that the user does not intend. In our previous works, the user has driven our assistive wheelchair in the usual manner. Our proposed system estimates its user’s intentions and passively works to complement their intentional force by negating the wheel traction that is generated by the road’s inclination using only the servo brakes on each wheel. Nevertheless, in some cases, our system fails to assist the driving motion of its user because the user drives the wheelchair in several ways that depend upon the environmental condition, for example, during uphill or downhill driving. The required assistance is not constant according to the situation, and it is difficult to assist with one wheel-control algorithm. Therefore, in this study, we first investigate the required assistance condition according to the driving situation by conducting a preliminary experiment with wheelchair users. Considering the results of this investigation, we then propose a novel user interface that intuitively shows the system information and a wheel-control algorithm that selects a suitable wheel controller according to the driving situation.

The preliminary announcement, for example, the turn-signal of the automobile, the track's announcement use turning or reversing and voice announcement of the train arrives at the platform are used in the various scenes in a daily life. These preliminary announcements are used to suppress the negative impression such as discomfort, aversion and surprise of the human. In this sturdy, we clarify the effective service robot's preliminary announcement that does not give discomfort and aversion to humans in the actual outdoor symbiotic environment. In this paper, we clarified the effective preliminary announcement of the service robot in a real outdoor symbiotic environment that is used three revolving lights on the top of our prototype robot. Furthermore, we investigated the effectiveness of preliminary announcement in the actual outdoor symbiotic environment.

In recent years, service robots have tended to have the ability to move autonomously in an environment in which they co-exist with humans. Thus, the robots must engage in obstacle avoidance and arrive at the goal point. The potential method is one of the path planning methods for obstacle avoidance. Because pedestrian is moving obstacle, he/she may go outside of artificial potential in a similar method suggested by a previous study. However, if the robot avoids collision too securely, robot's performance is limited so a trade-off problem arises between collision avoidance and robot's performance. A potential method requires a function that allows a robot to plan an effective path without collision with pedestrian by predicting a pedestrian's next position during robot operation. In this study, we develop a robot navigation method according to the characteristics of pedestrian flow. We spread a repulsive potential that adapts the pedestrian potential equation based on the route estimation model. In the proposed method, the repulsive potential changes depending on the position or the velocity of the pedestrian. The robot plans a path without including unnecessary potential, according to the pedestrian's route estimation range. Compared with the previous potential method, wherein the potential field is distributed equally around a pedestrian the same way as a static obstacle, the proposed method enables the robot to plan an effective path with pedestrian characteristics. The proposed method's effectiveness is demonstrated through experiments.

This paper proposes an automatic sitting posture coordination system for avoiding the risk of the pressure sore on a wheelchair. For avoiding this risk, it is important to maintain the suitable posture which reduces the pressure concentration and share stress between the buttock and a sitting surface on the wheelchair, and to depressurize them as the healthy person does during sitting on the wheelchair. For realizing these functions, this paper proposes an active seating support system which consists of a sitting surface lifting device and a backrest tension coordination device. Our system estimates the pressure concentration and share stress on the buttock and based on the estimated results, our system realizes the suitable sitting posture which minimizes them coordinating two devices. Furthermore, when the risk of the pressure sore increases, our system assists to depressurize the pressure concentration and share stress with this suitable sitting posture. Using our proposed active seating system, the patient can depressurize by a natural body motion as the healthy person automatically. The performance of our system is verified by experiments using our prototype.

A physical activity estimation scheme is proposed for patients who use a robot for standing assistance. In general, conventional assistive robots do not require patients to use their own physical strength to stand, which leads to decreased strength of the elderly. Therefore, an assistive robot that maximally uses a patient's remaining physical strength is desired. The assistive robots can achieve this objective by estimating the physical activity of the patient when they stand. The activity estimation proposed here is primarily based on a human musculoskeletal model of a lower limb, which exhibits a biarticular muscle function. The patient generates a natural standing motion using the biarticular muscle function, and the proposed model enables the assistive robot to estimate the patient's physical activity, without using biosensors, such as electromyographs, which are normally stuck on patients. The proposed estimation is implemented with a prototype assistive robot that assists elderly patients to use their remaining physical strength based on the estimated results, thus testing the effectiveness of the proposed method.

This paper propose a driving assistance scheme for a powered wheelchair with a force feedback function of its joystick. Generally, a powered wheelchair moves in a human coexistence environment such as a pedestrian way and it is important to move on these environments without a feeling of oppression to other people around it. Therefore, we propose a concept of a dynamic personal space which is a region surrounding a person who can feel comfortable psychologically to other person with a moving wheelchair in the pedestrian flows. A dynamic personal space is defined considering with a relative velocity and acceleration between the powered wheelchair and the pedestrians around it. Using this concept, we develop a driving assistance system using a force feedback of its joystick, which enables the powered wheelchair to move in a human coexistence environment without a feeling of oppression on other people. In this paper, we investigate spatial relationship with a powered wheelchair experimentally and define a dynamic personal space using these results. For verifying its effectiveness, the performance of our system is verified by experiments using our prototype.

This paper describes a localization technique for a mobile robot with unknown obstacles, such as pedestrians. It is important that a robot can localize in a crowded environment stably in order to perform navigation efficiently. However, In real environment, robot may be unable to run to a target position, because the robot cannot localize the self-position correctly with unknown objects, such as pedestrians. In this paper, this problem is solved by proposing the stable localization technique using road surface images in a real environment. Road surface images is able to be observed stably even in a crowded environment because it frees of influence of unknown objects, we verified the validity of the proposed method by the localization experiment and the goods delivery demonstration in real environment. From the experimental result, we checked that the proposed method can localize stably even in a crowded environment where unknown objects exist.

In recent years, the demand for service robots is increasing all over the world. Service robots have the purpose to assist living, such as medical treatment, transportation and security, and tend to have the function which move automatically in an environment. When operating such an autonomous mobile robot in a real world, the obstacle avoidance is the necessary technique to use a robot safely. A robot has to avoid static obstacles such as garden plants or signboards, and dynamic one such as pedestrians, and especially the robot has to act with caution to pedestrians. The purpose of this paper is analysis of human walking characteristics from the movement of a pedestrian, and from the experimental result, it turned out that the future movement of a pedestrian will be able to predict by movements of a few pedestrians.

This paper describes a localization technique for a mobile robot with unknown obstacles, such as pedestrians. It is important that a robot can run in environment freely in order to perform goods conveyance efficiently. In order to run in environment and deliver goods to a target position, self-position presumption is important. However, In real environment, robot may be unable to run to a target position, because the robot cannot localize the self-position correctly with unknown objects, such as people and goods. In this paper, this problem is solved by utilizing the robust position localization technique based on the observed space information to a delivery robot's navigation. Furthermore, by building the system which can set a robot's run path easily, flexible path creation and autonomously running are realized. we verified the validity of the proposed method by the localization experiment and the goods delivery demonstration in real environment. The system which localizes self-position with 0.1m error even in the environment where an unknown object exists, and performs goods delivery to a target position was realized.

This paper describes a localization technique for a mobile robot with unknown obstacles, such as pedestrians. Many robust localization technnique with unknown obstacles by using a particle filter and laser scanner have been propsed. In particular, Tomizawa's method(Space observation model) is effective. However, this method has problem that it does not consider the fixed object occlusion. Thus, in the environment where thin wall (this is example), the robot position cannot be localized well. In this research, we solve this problem by rebulding the space information which is observed intrinsically in predicted position by considerinmg the fixed object occlusion. From above, we realize robot position localization with more high accuracy than previous method, by solving the problem of previous method and improving the previous method. We evaluate the validity of the proposed method by a simulation and an experiment in a real environment.