Junji Satake

Multiple target tracking by appearance-based condensation tracker using structure information

Multiple target tracking is a challenging problem, especially when targets are frequently crossing each other. It becomes very difficult and confusing when some targets are often occluded by other targets. This paper proposes a novel tracking method for the problem using an appearance-based condensation tracker. In order to overcome difficulties in the occlusion problem, a target object is regarded as a set of parts that constrains each other in the target structure. While each part is tracked basically in the condensation method, all the parts cooperate in the drift step of the condensation. Experimental results show the effectiveness of the proposed method for multiple person tracking and human face tracking.

Person-independent face tracking based on dynamic AAM selection

We have developed a high-precision method that selects an appropriate model of a video image in order to track an unknown face in front of a large display. Currently, Active Appearance Models (AAMs) are used to track non-rigid objects, such as a faces, because the models efficiently learn the correlation between shape and texture. The problem with an AAM is that when it tracks an unknown face, excessive training data increases tracking errors because there is an intermediate model size beyond which the reduction in fitting performance outweighs the gains from any improved representational power of the model. To increases the accuracy with which an unknown face is tracked, we built clustered models from training datasets and select a cluster that includes a face which is similar to the unknown face. Our method of clustering and cluster selecting is based on the Mutual Subspace Method (MSM). We demonstrated the effectiveness of our method by using the leave-one-out cross-validation.

Stereo-Based Multi-person Tracking Using Overlapping Silhouette Templates

This paper describes a stereo-based person tracking method for a person following robot. Many previous works on person tracking use laser range finders which can provide very accurate range measurements. Stereo-based systems have also been popular, but most of them are not used for controlling a real robot. We previously developed a tracking method which uses depth templates of person shape applied to a dense depth image. The method, however, sometimes failed when complex occlusions occurred. In this paper, we propose an accurate, stable tracking method using overlapping silhouette templates which consider how persons overlap in the image. Experimental results show the effectiveness of the proposed method.

A Real-life Test of Face Recognition System for Dialogue Interface Robot in Ubiquitous Environments

This paper discusses a face recognition system for a dialogue interface robot that really works in ubiquitous environments and reports an experimental result of real-life test in a ubiquitous environment. While a central module of the face recognition system is composed of the decomposed eigenface method, the system also includes a special face detection module and the face registration module. Since face recognition should work on images captured by a camera equipped on the interface robot, all the methods are tuned for the interface robot. The face detection and recognition modules accomplish robust face detection and recognition when one of the registered users is talking to the robot. Some interesting results are reported with careful analysis of a sufficient real-life experiment

Human-robot interaction in the home ubiquitous network environment

The situation recognition ability of the robot is enhanced by connecting a home ubiquitous network and conversational robots. The situation explanation ability of the robot is also enhanced by acquiring information through the network. The new development of the human robot interaction can be expected in total. In this paper we describe a prototype system that is developed in the experimental house based on such a concept. Then, the study of the actual proof experimental life in the house is discussed.

Autonomous monitoring framework with fallen person pose estimation and vital sign detection

This paper describes a monitoring system based on the cooperation of a surveillance sensor and a mobile robot. Using a depth camera which acts as the surveillance sensor, the system estimates the pose and orientation of a person utilizing a skeleton-based algorithm. When the person fell down, the sensor sends the persons pose and orientation information to the mobile robot. The robot determines the possible movements and strategies for reaching the fallen person. The robot then approaches the person and checks the vital condition whether the person is breathing, and the recognition result is notified to a hand-held device. Experiments on our monitoring system confirm a successful series of the autonomous operations.

An RT Component for Simulating People Movement in Public Space and its Application to Robot Motion Planner Development

This paper describes a software module for simulating “people movement” in public space such as shopping centers and cafeterias. We decompose people movement into global and local, and make a model of each of them. Global movement corresponds to following a route from a current position to a destination. In local movement, a person moves toward the next subgoal while avoiding surrounding persons and obstacles. We also model behavior specific to a cafeteria, such as queuing and searching for unoccupied seats. We implement these simulation algorithms in a simulator RT component, that can be used easily for development of robot motion planners, which are also realized as RT components. Various simulation experiments show the effectiveness of the simulation algorithms and the simulator RT component.

Stereo-based road boundary tracking for mobile robot navigation

This paper describes a method of stereo-based road boundary tracking for mobile robot navigation. Since sensory evidence for road boundaries might change from place to place, we cannot depend on a single cue but have to use multiple sensory features. The method uses color, edge, and height information obtained from a single stereo camera. To cope with a variety of road types and shapes and that of their changes, we adopt a particle filter in which road boundary hypotheses are represented by particles. The proposed method has been tested in various road scenes and conditions, and verified to be effective for autonomous driving of a mobile robot.

A fast stereo-based multi-person tracking using an approximated likelihood map for overlapping silhouette templates

This paper describes a method of tracking multiple persons with occlusions using stereo. Many previous stereo-based systems track each person separately and do not explicitly handle such occlusions. We previously developed an accurate, stable tracking method using overlapping silhouette templates which considers how persons overlap in the image. However, because the method uses a particle filter, a lot of processing time is needed for estimating each particles likelihood by comparing many templates with the image. In this paper, we propose a new method which can decrease the number of image comparison by using an approximated likelihood map based on kernel density estimation. Experimental results show that the proposed method is able to reduce the processing time greatly without dropping the tracking performance.

A hierarchical SLAM for uncertain range data

This paper describes a new approach to SLAM problems using low quality range data. Vision sensors are useful for acquiring various kinds of environmental information but range data obtained by stereo vision is less reliable than other active sensors like laser range finders. False stereo matches often result in spurious obstacles, which may degrade the map when directly used in existing SLAM methods. We therefore propose a hierarchical approach in which local probabilistic occupancy maps are first generated to filter out such spurious obstacles and then used as inputs to an RBPF-based SLAM. Experimental results in simulation and in a real environment show that a consistent map can be generated by the proposed method with low quality stereo range data.