This paper describes a method for tracking of moving objects in multiple categories with a LRF (laser range finder). The proposed method can track multiple moving objects robustly using a framework based on SJPDAF (sample-based joint probabilistic data association filter). It can continue tracking correctly if moving objects get close to other moving objects, and can track moving objects in different categories at the same time. We implemented the proposed method and obtained experimental results demonstrating its effectiveness.

This paper describes environmentmaps that are comprised of the following three types of information, 1) 3D environmental changes that represents human activities, 2) human trajectories in 2D that represent how humans move in the environment, and 3) human posture data. These maps are utilized in order to plan safer, quicker and/or non-human-disturbingpaths for a mobile service robot at themuseum “Miraikan”. Experiments are conducted within “Miraikan” and results are shown.

Localisation and mapping are fundamental capabilities for autonomous mobile robots, and there has been a large amount of recent work in these fields. However, much of the work does not consider dynamic environments that include humans and moving objects. Such objects can cause occlusions resulting in a fewer visible landmarks, which can decrease localisation performance. This paper describes a novel method of localisation and multi-layered 2D mapping in dynamic environments using selective updating of a particle filter. A number of horizontal, planar laser scans at varying heights are used to construct a number of corresponding 2D maps. At each mapping step, the position estimate from the map layer which minimizes uncertainty is selected and used to update all maps. Using the proposed method, it is possible to localize precisely in dynamic environments, despite the effects of occlusion. Experimental results in a large outdoor environment confirms the effectiveness of the method.

Recently robotic localization and mapping have been a field of study. They are the basic function of a mobile robot. However such methods are not enough in the dynamic environment including human and moving objects. In such field, there are serious problems such as occlusion and fewer landmarks, that make localization error. This paper describes a nobel method of localization and multi-layered 2D mapping in the dynamic environment using selective update for particle filter. It is possible to localize precisely, in spite of negative factor in the dynamic environment. Also we can exactly make multi-layered 2D maps.

It is important for robots to acquire the information of a person's grasped objects when robots assist or communicate with the person. Kinect can extract the image of a person and his/her grasped objects, therefore to extract the image of the grasped objects is the problem to be solved. To perform image extraction for such subjects, it had been proposed methods using Image Feature Extraction and Machine Learning. However, these require a huge learning image data. Therefore, this study proposes the method with Kinect by using Body Dimension Database. By performing the Multiple Regression Analysis, to obtain wrist position, We estimated arm lengths, which is unable to measure adequately with Kinect, with human body measurements which is able to measure adequately with Kinect. And we acquired the information of a person's grasped objects by extracting the image of hand grasped objects.

Robotic mapping such as two-or three dimensional occupancy grid map has been a field of study. However such occupancy maps are not enough for advanced tasks including object search, object grasping and change detection. We focus on tables and tabletop objects because a table is one of the most important tool in human activity. This paper describes an observation system of tabletop objects based on 3D-sensors and environmental map. We developed a mobile robot equipped with a Velodyne HDL-32E Lidar sensor and a Microsoft Kinect RGB-D sensor.

This paper describes an autonomous vehicle navigation system based on Velodyne LIDAR. The system mainly focusing on an autonomy at the car park, and it includes following functions, 1) odometory correction, 2) 3D map building, 3) localization, 4) detecting human trajectories as well as static obstacles, 5) path planning, and 6) vehicle control to a given trajectory. All those functions are developed on ROS. Car park of 70x50[m] area is used for experiment and results are shown.

This paper proposes a method to segment out a hand grasped object from human region obtained from Kinect sensor by using body dimension database. Having dataset of human body dimensions, Multiple Regression Analysis is applied to find out the best explanatory variables for forearm and upper arm length. As a result, "body height" is selected. In order to measure "body height" accurately, Kinect depth image is utilized to search with kinematical result obtained from Kinect software. After estimating wrist position, we can segment out hand grasped region. Methods and experimental results are shown.

It is important for service robots to recognize human activity. A table is the center of human life. Objects on the table are changed by human. Therefore service robots need to find objects on the table and detect changes. This paper describes an observation system of tabletop objects. Our system includes a perceptual pipeline of RGB-D point cloud and a planning of the location to observe a table. In addition, the location of objects, and their identities are recorded in the database. This system is useful for any application that involves dealing with objects, including grasping, change detection, and object search. We demonstrate a robot equipped with a Microsoft Kinect RGB-D sensor and a Velodyne HDL-32E Lidar sensor build on our system.

This paper presents a service oriented map that includes a) environment shape, b) human walking trajectory, and c) human sitting/standing position. Such information is important for a robot that provides services to human beings. Environmental shape (a) and human walking trajectory (b) are obtained by LRF. Human sitting/standing position (c) is obtained from Microsoft Kinect sensor by actively positioning and orienting the sensor to confirm human posture. A combined algorithm to obtain an integrated map is described, and experimental results follow.

A service robot can provide the advanced service suitable for the person by not only recognizing people, but also identifying and classifying people. Therefore, in this study, a classification of people and identification function are realized by computing the amount of the features which expresses a form from people's size. I performed a multiple regression analysis from the measure of human body which can be presumed from KinectSDK to the "individual feature quantity" extracted from measure of human body data using principal component analysis. As a result, it was shown that the first main ingredients of individual feature quantity can be expressed by the contribution of 88.9%, and the second main ingredients can be expressed by the contribution of 76.1%.

This paper describes a method to estimate a set of body dimensions of a subject using the Microsoft KINECT sensor. Principal Component Analysis(PCA) of the AIST anthropometric database shows first and second principal components are interpreted as scale and degree of obesity respectively. A few body dimensions were chosen by linear multiple regression analysis. Applying the resulting estimator to a given subject using the KINECT sensor to obtain joint positions and 3 body dimensions by analyzing depth image, we can estimate a set of body dimensions for the subject. The system is described and experimental results are shown.

Environment and human recognition is necessary for service robots, such as guidance, communication, or guard. Therefore the technique is suggested that record the human positional information obtained by sensors as map information. By adding human posture information to this map information, robots will be able to offer service depending on human position and posture. Therefore, in this paper, by using Kinect sensor produced by Microsoft, we judge human sitting or standing posture that people take frequently. And we suggest unification method of two kinds of map information, 1) human trajectory, and 2) sitting or standing position.

Path planning uses map together with its cost to plan a path that contains minimum cost. Usually Euclidian distance is used for map cost. In this paper, we propose two kind of maps, 1) 3D obstacle map and 2) human usage of free area. Observing human trajectories, and clustering their histories by modified k-means method, a) working area, b) pedestrian area, and c) pedestrian crossing area. Using those information, application to robot navigation is also shown.