|
|
Activities
Reports-->
Research
Activities of Hashimoto Lab
Abstract:
This document introduces the research
activities of Hashimoto Lab, Institute of Industrial Science,
The University of Tokyo. The research activities are categorized
in three main projects: Intelligent Space Project (iSpace),
Intelligent Transportation Systems Project (ITS), and
Micro- Telemanipulation Project.
Often, science fiction movies become
good references for actual engineering. With the progress
of technology, some of these previously unbelievable systems
appeared first in science fiction literature have actually
become a part of everyday life like for example space rockets
and robots. In our laboratory we are developing such a system,
but before explaining exactly what it is, I would like to
continue with some thoughts about a high intelligence computer
named HAL from the movie Space Odyssey 2001. HAL can watch
a human's activity with its distributed cameras and control
subordinate systems as expanded actuators of it. In another
SF movie, titled 'Demon Seed', a brilliant scientist of
the future creates a computer named 'Proteus' with almost
limitless intelligence. However, Proteus tried to produce
offspring and it hinders all the people who plan to get
rid of the Proteus. As HAL did, Proteus utilizes all the
electrical systems in the house as its parts. Unfortunately
both of the movies are telling us to fear technology when
the machine gains high intelligence. However, we should
notice the intelligent systems in those movies. Some people
may say these are ubiquitous computing, but we recognized
those systems as an intelligent environment. Such intelligent
environments are able to watch what is happening in them,
build a model of them, communicate with their inhabitants
and act based on decisions they make. Especially the capability
of the environment to act as a context-sensitive user interface
(e.g. to respond to gestures) and react in certain situations
(e.g. accidents, intruders) promises a range of application
scenarios such as intelligent hospital rooms, office, factory,
asylum for the aged, etc. Research should focus on intelligent
man-machine systems, which resemble a welfare support system.
Project Members:
| Kazuyuki Morioka |
PhD student |
| Péter Szemes |
PhD student |
| Machiko Chikano |
Researcher, Yamatake Co. |
| Noriaki Ando |
National Institute of Advanced Industrial
Science and Technology |
| Joo-Ho Lee |
Tokyo University of Science |
| Yoshihiro Yamashita |
PhD student |
| Hiruyuki Isu |
Master Student |
| Hwang Gil Gueng |
Master Student |
| Sosuke Takatsuka |
Master Student |
| Yoichi Kuroda |
Research Student |
| Mihoko Niitsuma |
Research Student |
Hashimoto Lab. in University of Tokyo
has proposed Intelligent Space since 1996. At the beginning
it was consists of two sets of vision cameras and computers
with a lab-made 3D tracking software which, was written
in C and TCL/TK under Linux. Later, large sized video projector
(100 inches) was added to the Intelligent Space as a display/actuator.
Mobile robots were introduced in the Intelligent Space to
support inhabitants as well as for being supported. Vision
cameras and computer sets were arranged to corner the entire
room and formed the Intelligent Space.
Making space intelligent can be defined
as a space with functions that can provide appropriate services
for human beings by capturing events in the space and by
utilizing the information intelligently with computers and
robots.
Robots with partial intelligence become
more intelligent through interaction with the space. Moreover,
robots can understand the requests (e.g. gestures) from
people, so that the robots and the space can support people
effectively. This space using intelligence is called the
Intelligent Space. The concept of the Intelligent Space
is shown in Figure 1
DIND (Distributed Intelligent Network Device) has a sensing
function through devices such as a camera and microphone
that are networked to process the information in the Intelligent
Space.
The Intelligent Space can physically and mentally support
people using robot and VR technologies; thereby providing
satisfaction for people. These functions will be an indispensable
technology in the coming intelligence consumption society.
Figure 1: The Intelligent Space.
|
Requiring Functions
On the software side, we have three different types of
tasks with different characteristics. They are distributed
as individual processes over the computers of the network.
- Sensor and Actuator Servers
For the data preprocessing highly specialized modules
are needed that derive relevant information from the sensors
and offer this information on the network.
- Intermediate Processing
On an intermediate level process collect data from one
or several servers to which they connect as a client.
Typical task are sensor fusion, temporal integration,
and model building. As sometimes this requires some real-time
capability, they should be located close to the sensor
computers. The intermediate results are again offered
on the network.
- Application Process These processes perform the actual
applications of the space. As they usually require low
volumes of data and slower reaction times optimization
is less critical. They should however be easily portable
across architectures and easily maintainable by the user.
Distributed Intelligent Network Device (DIND)
In the Intelligent Space, the DIND understands
events in the space and provides appropriate services for
people by using devices such as robots, displays, and speakers.
A DIND is composed of sensors, a processor for the information
from the sensors, a network for information interchange,
and a power source. It is microminiaturized and a low-cost
device. Since networking is a prerequisite, functions such
as a high level of security, self-diagnosis, and function
sharing are essential [2]. Figure 2 shows
the concept of DIND. As constructed, DIND needs MEMS (Micro
Electro Mechanical System) and nanotechnology for its miniaturization.
Figure 2: The concept of DIND.
|
Each DIND inherently has an intelligence limit because
of its size; however networking decentralized DINDs in the
space can realize a high level of intelligence through their
autonomous cooperation. The basic functions of networked
DIND are as follows:
- Observation of events in the space
A function to observe events in the space via sensors
such as visual, infrared, radio frequency and ultrasonic
sensors, high-sensitivity microphone, and laser radar.
- High level processing of the obtained data
A function to process locally obtained data and to output
results to the network in a sensor-independent format.
- Intelligent decision
A function to suppose the events in the space by utilizing
information from networked DINDs and past data, and to
make an appropriate decision (a cooperative activity in
networked DINDs).
- Offering of appropriate services
A function to issue commands to robots and/or manipulators
for physical support.
The intelligence, placed spatially by DINDs, connects
the physical and digital spaces, and realizes the understandings
of people's intentions and the appropriate services for
them. The Intelligent Space is a place where various technologies
merge into one on a DIND base, and thereby evolves as a
platform. Figure 3 shows
the concept of a network system. The DIND network connects
outdoor and indoor spaces seamlessly, forming a large Intelligent
Space. Various and diverse data such as family healthcare,
indoor child monitoring and out door traffic monitoring
are shared and/or processed by the Intelligent Space as
a platform to realize a wide variety of services.
Figure 3: DIND network system.
|
3D positioning of Human
To support humans in the space, the Intelligent
Space tracks humans. Recognizing the human is done in two
steps. First, the area or shape of a human is separated
from the background. Second features of the human as head,
hands, feet, eyes etc. are located. Taking the images of
several cameras, we can then calculate the 3D position of
the human. To calculate 3D from several camera views point
correspondences are needed. To establish these correspondences
directly from the shape of the human is difficult. Instead
we first find the head and hands of the human and use their
centers for matching. A second motivation to further analyze
the shape is that adaptive background separation in complex
scenes detects recently displaced objects. The above algorithms
are implemented in three different software modules (Camera
Server, 3D Reconstruction Module, Calibration Client) of
the Intelligent Space.
Map Building by Looking People
Mobile robots need maps of their environment
for navigation, localization and task specification. Mobile
robots can navigate robustly without a precise geometrical
model if some other way of localization is given and a topological
map is supplied. The approach we suggest is to look at the
movements of people in the room. In indoor environments
people and robots consider similar things as obstacles.
This method has the additional advantage that it detects
obstacles that most sensors fail to notice. Examples are
trapdoors, yellow lines on the floor or signs saying "Danger
- Don't Enter". Positions of moving persons were obtained
with about 20 Hz. Only positions with a vertical height
between 1.65 and 2.00 meters and only blobs with at least
0.6 times the size of a head were taken into account.
Mobile Robot Localization
To locate mobile robots in the Intelligent
Space, four colored spherical targets are placed around
the body of each mobile robot, as shown in Figure 4.
These targets are found by using the same technique as locating
human hands and heads. However as matching of several identical
blobs are difficult, we need a further clue to identify
matching targets. For this, we use color bar codes that
are located under the targets. The color bar code consists
of several colored fields. Each of the fields can have one
of the eight possible colors. With this technique, we are
able to distinguish the targets of mobile robots [3]. From this, the position and the orientation of
the robot can easily be derived (Figure 5).
As all targets have a known height, we are additionally
able to check for incorrect reconstructions. The mobile
robot gets information of its absolute position and angle
from the intelligent space through wireless LAN. However,
the mobile robot moves during the interval of communication
and interpolation by mobile robot's using internal sensor
are needed. We use simple dead reckoning algorithm only
with rotary encoders. Slip between wheels and floor is not
considered in dead reckoning. Due to the short interval
before the position correction from the intelligent space,
the error of the position and angle remains small. Positioning
error depends on the disposition of sensors, resolution
of sensors, distance between a robot and a sensor. In our
experiments, we verified a DIND, which covers 3m X 3m square
area, estimates pose of a robot less than 0.1m position
error and 1.5 degree directional error.
Figure 4: Pose estimation based on color
bar codes.
|
Figure 5: Deployment of color bar codes.
|
Human Machine Interface
Our life, the surrounding electronics
devices become more complex. There is a very important layer
between the User and the core of the machines. The aim of
this layer is to map between the human nature and the core
technology. In other words, this layer is the interface
between the human and the machine, that's why the name,
Human Machine Interface (HMI).
When the User exists in the Intelligent
Space (iSpace), intend to act natural way, give commands
with his/her body and his/her voice and also receive information
with his/her eyes, or via contact sensing. To map this natural
human activity into machine understandable format, complex
Human Machine Interface is required. The interface is complex
toward the system core, but easy to use face to the user.
The aim of this research to create a Human Machine Interface,
where the interaction between the Intelligent Space and
the User realized by three communication channels: Audio,
Visual and Haptic
The User enters into the iSpace with an
intention in his/her mind. The iSpace offers a lot of services,
so the user could choice the most suitable for his/her intention.
The service is activated via the HMI. The service allocates
the HMI resources to communicate with the user. The service
and user communication is realized with events and messages.
But the type form of the messages are different both at
the user's and the service's side. Generally, the service
is represented as a software what accepts software objects
as a message, but the user use physical gesture, for example
for communication. The translation is done by the HMI.
Our Human Machine Interface is designed
for personal communication between the inhabitant user and
the services of the iSpace. The HMI carries sensors and
actuators for audio, visual and haptic communication. The
sensors and actuators are connected to gesture and speech
recognition modules what translates the user's messages
to software messages. And opposite direction, computer graphics
system and speech synthesizer translates the software messages
into human messages.
Figure 6: Deployment of color bar codes.
|
Joint Research with:
| Péter Korondi |
Budapest University of Technology and
Economics, Hungary |
| Florin Dragan |
Politehnica University of Timisoara,
Romania |
| Emil Voisan |
Politehnica University of Timisoara,
Romania |
Human controlled direct training of an
adaptive system is very efficient in specifying complex
system behaviors. One of the main difficulties of training
a real system is the necessity of the real training environment
and the system itself. The training environment must be
a common model of all possible future situations, so to
built it only for the training process, hence, is not only
unnecessarily expensive (considering the possible real damages
can be appeared during the failures of the training process),
but in some cases unsolvable. Another aspect of the necessity
of virtual workplace is already appeared in telerobotics.
Many application areas are unsafe for the human operator
(trainer in this case), or simply out of the human sized
world, or the human living environment, such as space, undersea,
or medical micro operations. To solve these difficulties,
we suggest adopting the idea of virtual workplace (known
from telerobotics) for training: the virtual training.
Similarly to telerobotics, the virtual workplace is a humanized
model of the real workplace. Having the virtual workplace,
the operator can perform the necessary training situations
in a human sized safe environment. In telerobotics, the
virtual workplace is normally connected to a real system,
which performs the real operations based on the operator's
remote commands. In our case this system is practically
a simulated model of the training environment and the system
to be trained. The main benefits of this structure beyond
its low expenses are its safety and flexibility. Having
a simulated system and training environment there are no
real damages can caused by operator errors or control action
error, moreover the reconfiguration of the training environment
can be done very quick and simply.
The aim of Virtual Room (VR) research
project is to recreate an environment of a physical experimental
space for studying different motion control and vision algorithms
for a given robot before real world implementation. The
present virtual space is the recreation of the Experimental
Intelligent Space of Hashimoto Lab at the University of
Tokyo. The room currently contains the following objects
- Passive objects: desks, chairs
- Active objects: robot agents, like Mobile Robot Assistant
- Sensors: CCD cameras
- Actuators: Large Screen
The project is developed in C++, and graphical
implementation of the objects is achieved using Coin/Open
Inventor library. Inventor's foundation is supplied by OpenGL
and UNIX, Inventor represents an object-oriented application
policy built on top of OpenGL, providing a programming model
and user interface for OpenGL programs. The current development
operating system is a Suse Linux 8.1, and Coin3D version
is 1.0.4. The present state of the Virtual Room includes
graphical representations of the objects mentioned above.
The graphical environment allows a walk through the virtual
space and it is also possible to visualize the virtual image
of each camera with this configuration. The images of a
virtual camera and a real camera are compared in Fig. 7.
. Both rooms (virtual and real) have 8 pan-tilt-zoom cameras,
and one more is mounted on the Mobile Assistant Robot.
Figure 7: Virtual Room (upper) and the
Experimental Space (lower)
|
| Massaki Wada |
PhD student |
| Xu-Chu Mao |
PhD student |
| SungSik Kim |
Master Student |
Theory and application of stochastic
algorithms have also been investigated in our laboratory.
Research has been conducted mainly in the field of nonlinear
estimation algorithms and their applications to real systems.
Theoretical studies main motivation has been the development
of algorithms that would enable increase system designers
"expressive power" (use of available knowledge about a system,
including complex mathematical models, and incomplete empirical
system features knowledge). The algorithms would allow to
make full use of recent years ever increasing computational
power, increased data collection capabilities of advanced
sensing systems, and abundant data gathering capability
(e.g. by networked systems). The developed algorithms are
expected to have wide applicability, being useful in any
system that would benefit from increase in inference "precision".
Research on these algorithms application, and development
of advanced algorithms for particular systems has also been
actively conducted. Our investigations have been related
with automotive systems. Recently, we have focused in the
development of advanced signal processing algorithms for
GPS (global positioning system).
A modeling and estimation framework including
learning and selection of nonlinear dynamical systems model
have been developed/proposed. An essential contribution
of this work is the proposal/development of a new nonlinear
dynamical systems parameter learning algorithm. With the
present state of technology, there are many systems where
it is possible to measure or collect significant amount
data, and also derive a mathematical model for the system.
The proposed framework would enable increase in the application
domain of estimation algorithms allowing the use of these
data to learn such system parameters (including the system
noises parameters), and "complete" the model that may be
used in inference algorithms. Various filtering algorithms
(named generically as particle filters) for non-linear,
non-gaussian model states estimation have been recently
investigated. These filters allow inference of more "general"
structure system. However, in the case of high dimension
models, it is difficult to realize real-time filtering.
Another theoretical work has proposed/investigated a new
Rao-Blackwellisation particle filter based algorithm for
real-time applications.
Despite of its increased use, GPS still
has many limitations that restrict its use in many systems.
It was also verified that despite the continuous development
of modern receivers, most of GPS receivers still rely on
simple signal processing algorithms. So, our approach in
the research on GPS has applied modern signal processing
algorithms to develop improved performance receivers. Firstly,
a new algorithm for state estimation of standalone GPS that
fuses two different GPS measurements" (pseudorange and Doppler
shift) was proposed and developed. Following research is
investigating the development of a weak signal GPS that
would allow improved urban and indoor navigation. The following
subsections will further describe these researches.
This research aims to improve the precision
and robustness for GPS, mainly in bad measurement conditions.
We developed a new GPS algorithm for standalone positioning
that fuses pseudo-range and Doppler shift measurements using
a more complete system state model. The pseudorange is the
C/A code cycles and phase between the satellite and receiver
(figure 8), and
Doppler shift is the apparent change in the frequency of
a carrier signal caused by the relative motion of the satellite
and receiver (figure 9).
Figure 8: Pseudorange Measurement Model
|
Figure 9: Doppler Shift Measurement Model
|
The main contributions of this work are:
the proposal of a process model including pseudorange non-white
error and clock error, and its use in the unscented Kalman
filter (UKF). Comparing with the past conventional positioning
methods, the experimental results showed that the proposed
method has better performance with higher precision and
robustness.
By now, positioning in outdoors with
clear view of the sky has been raised to provide high-level
measurements. However, in the urban canyon, tunnel and indoors,
GPS signals are too weak to be processed by conventional
GPS receiver. We proposed a new GPS receiver structure for
weak signal processing, it's aim is to perform real-time
3-D GPS positioning in such bad conditions. The main steps
of GPS signal processing are signal acquisition and tracking.
Conventional GPS receiver uses simple algorithms for implementing
signal acquisition and tracking, and can't perform weak
signal processing.
Figure 10: Acquisition using FFT
|
Software receiver uses FFT algorithm (Figure
10) to perform correlation
calculation and is able to perform weak signal acquisition
using block-processing techniques. However, software receiver
has complex structure, and can hardly perform real-time
weak signal processing. We think that advanced signal processing
algorithms may be used to partially solve weak signal related
signal processing problems. So, we have proposed a weak
signal GPS receiver architecture based on advanced signal
processing techniques (Figure 11).
Figure 11: Proposed GPS Receiver architecture
|
GPS signal tracking includes carrier tracking
and code tracking. The proposed architecture use hardware
correlators (used in conventional receivers) to retain fast
processing and high-resolution of code phase measurements.
The use of Kalman filter in the signal tracking algorithm
is proposed to improve the tracking performance and robustness.
We are also investigating a new acquisition algorithm for
use in the system. Till now we develop a combined carrier
tracking and code tracking models using an Unscented Kalman
filter. Comparing with the standard tracking method, simulation
results showed that the proposed algorithm could track weaker
signal than conventional methods.
Project Members:
| Noriaki Ando |
National Institute of Advanced Industrial
Science and Technology |
| Péter Szemes |
PhD student |
| Hwang Gil Gueng |
Master Student |
In recent years, the interest in multi-modal
based collaboration systems has been continuously increasing.
This is also known as CSCW (Computer Supported Collaborative
Work) technology, which realizes easier collaboration among
distributed work groups. However little attention has been
given to the multimodal collaboration with physical contact.
Besides, optical/electronic parts and
components for new electrical devices, (including mobile
PC, PDA, etc.) are becoming smaller and smaller. Consequently,
there is an increase in the small-scale manipulation exemplified
by the mechanical lenses driver alignment in CD/DVD player
and small printed circuit board repair work. Moreover, there
is an emergent necessity of collaboration tools in industry,
which allows a better connection between laboratories, offices
and factories. We have developed tele-micromanipulation
systems with haptic feedback to support collaborative work.
Micromanipulation systems have also been studied briskly
in recent years, and application to biotechnology field,
medical field, nanotechnology field, among others is expected.
Control schemes for master haptic interface of tele-micromanipulation
systems is the main topic of this research [20].
In micromanipulation, visual information
of microenvironment is usually caught by microscope. It
is difficult to manipulate micro objects for human operator
based on single visual information. Getting 3D geometry
information of microenvironment is indispensable to human
dexterous manipulation. However space and cost problems
make difficult to use two or more microscopes, and they
still do not provide enough information for human dexterous
manipulation. In manipulation, the haptic feedback is very
important for human operator, and it is the reason why haptic
interface is adopted to our tele-micromanipulation systems.
Figure 12
shows the configuration of our tele-micromanipulation system.
In these systems, the master input device used by the human
operator is called haptic interface. The slave manipulators
used directly to perform manipulation are called manipulators.
Our bilateral teleoperation system is realized using PHANTOM(TM)
( a commercial haptic interface ) or a joystick type haptic
interface specially developed for tele-micromanipulation
[21,22]. A parallel manipulator with an originally
developed mechanism is used as slave manipulator in the
teleoperation system. The slave manipulator and master device
systems are connected using Ethernet and they are used to
perform teleoperation through the network.
Figure 12: Micromanipulation Systems
|
6 D.O.F. parallel mechanisms is used as
the slave manipulator. In general, parallel link structure
has good characteristics of precision and stiffness, although
it has small workspace and many singular points. We selected
parallel manipulator as our slave manipulator, because these
features are advantageous for micromanipulation. The details
of parallel manipulator are not discussed deeply here for
lack of space.
In this system, the 6 D.O.F. haptic device,
which was newly developed, is used as the master input device
(Figure 13). A
serial link mechanism is adopted using the linear motors
in the master device to realize large workspace in a compact
way. Three linear servomotors are used to perform translational
motion of X-axis, Y-axis and Z-axis.
Rotational motion is performed through
three AC servomotors installed orthogonally in the Z-axis
linear motor. A 6-axis force/torque sensor is installed
on the yaw angle motor, and a joystick like grip is settled
on the force/torque sensor.
This master device workspace dimensions
are  in translational motion and,
 in rotational motion. The Real Time Linux (RTLinux) is used
as the operating system to perform 2.5 KHz sampling time
necessary for motion control. Input-output using motor,
rotary encoder, force sensor are performed using AD, DA,
counter, DIO boards connected to an extended bus.
Figure 13: Appearance and Structure of
a Master Device
|
4.3 Model Reference Adaptive Controller for Master Device
This master manipulator has the advantage
that there is little interference between axes because the
linear motors arrangement (each is perpendicular the others).
On the other hand, since rail sides of the linear motors
of Y-axis and Z-axis are perpendicular, there is nonlinear
variant friction between the rail and the slider depending
on the positions. Consequently even if equal force were
inputted to the master device in X-axis, Y-axis and Z-axis,
the response to the input of each axis changes with conditions
such as a performance of actuators, position dependent friction
force and system inertia. Therefore, the isotropy of the
response is spoiled. In order to realize haptic interface
with few burdens to operators and with natural response,
it is necessary to respond equally to all directions independent
of its condition.
To solve these problems, a control scheme
based on reference model following was introduced. Model
reference adaptive control (MRAC), model following control,
model matching control, and sliding mode observer based
method [23] have been proposed in the literature for
implementing reference model following control.
Among these control schemes, the exact
model of a plant is required for model matching control
and model following control. On the other hand, since MRAC
could be adapted also for an unknown plant, we applied MRAC
to the master haptic interface.
MRAC system may be realized using the control block diagram
shown in Figure 14.
Figure 14: Model Reference Adaptive Control
|
-
- 1
-
Guido Appenzeller, Joo-Ho Lee, Hideki Hashimoto, ``Building
Topological Maps by Looking at People: An Example of Cooperation
between Intelligent Spaces and Robots'', IEEE/RSJ International
Conference on Intelli-gent Robots and Systems, pp. 1326-1333,
1997.
- 2
-
Hideki Hashimoto, Naoto Kobayashi, Toru Yamaguchi, ``Intelligent
Interactive Space'' (in Japanese)'', Lecture of the Japan
Society of Mechanical Engineering, robotics and mechatronics,
2BI4, 1998.
- 3
-
Joo-Ho Lee, Guido Appenzeller, Hideki Hashimoto, ``A Physical
Agent for Intelligent Spaces: Functions and Roles of Mobile
robots in Sensored, Networked, Thinking Spaces'', IEEE
Conference on Intelligent Transportation Systems, 1997.
- 4
-
Joo-Ho Lee, Hideki Hashimoto, ``Global Positioning Sys-tem
for Mobile Robots with Distributed Sensors'', IEEE/RSJ
International Conference on Intelligent Robots and Systems,
1999.
- 5
-
Joo-Ho Lee, Takahiro Yamaguchi and Hideki Hashimoto, ``Human
Comprehension in Intelligent Space'', IFAC Conference
on Mechatronic Systems, Vol.3, pp.1091-1096, 2000.
- 6
-
Kimihiko Nakatsukasa, Joo-Ho Lee, Hideki Hashimoto, ``Detecting
bewilderment from natural movement of peo-ple'', JSME
Annual Conference on Robotics and Mecha-tronics, pp. 2CII4-2(1) (2),
1998.
- 7
-
Kazuyuki Morioka, Joo-Ho Lee, Hideki Hashimoto, ``Mobile
Robot Control for Human Following in Intelligent Space'',
International Conference on Control, Automation and Systems,
2001. Hideki Hashimoto, Joo-Ho Lee, Kazuyuki Morioka,
``Human Robot Interaction via Intelligent Space'', Proceedings
of the 2002 International Conference on Control, Automation
and Systems (ICCAS2002), pp.512-517, 2002.10
- 8
- Joo-Ho Lee, Takashi Akiyama, Hideki Hashimoto, ``Study
on Optimal Camera Arrangement for Positioning People in
Intelligent Space'', Proceedings of 2002 IEEE/RSJ International
Conference on Intelligent Robots and Systems, pp.220-225,
September 30- October 4, 2002
- 9
- Hideki Hashimoto, ``Intelligent Space -How to Make
Spaces Intelligent by using DIND?-'',Proceedings of IEEE
International Conference on Systems, Man and Cybernetics
(SMC'02), 2002.10
- 10
- Joo-Ho Lee, Kazuyuki Morioka, Hideki Hashimoto, ``Physical
Distance based Human Robot Interaction in Intelligent
Space'', Proceedings of the 28th Annual Conference of
the IEEE Industrial Electronics Society, pp.-, 2002.11
- 11
- Kazuyuki Morioka, Noriaki Ando, Joo-Ho Lee, Hideki
Hashimoto, ``Robust tracking of multiple objects using
color histogram in intelligent environment'', IEEE/ASME
International Conference on Advanced Intelligent Mechatronics
(AIM 2003), pp.533-538 , 2003.7
- 12
- Joo-Ho Lee, Kazuyuki Morioka, Noriaki Ando, Hideki
Hashimoto, ``Condition-based Placement of Distributed
Active Vision Sensors for Guiding Robots in Intelligent
Envirionment'', IEEE/ASME International Conference on
Advanced Intelligent Mechatronics (AIM 2003), pp.546-551
, 2003.7
- 13
- Noriaki Ando, Joo-Ho Lee, Hideki Hashimoto, ``Cluster-Camera
Networking and Geometric Configulation for Intelligent
Space'', IEEE/ASME International Conference on Advanced
Intelligent Mechatronics (AIM 2003), pp.521-526, 2003.7
- 14
- Peter T. Szemes, Florin Dragan, Emil Voisan, and Hideki
Hashimoto, ``Evaluation of Inhabitant's Walking Habit
in Intelligent Space'' Proceeding of IEEE/SICE Annual
Conference of the IEEE Industrial Electronics Society,
Hotel Roanoke and Conference Center, Roanoke Virginia,
USA, Nov. 2-6, 2003
- 15
- Peter T. Szemes, Joo-Ho Lee, Hideki Hashimoto, and
Peter Korondi, ``Guiding Assistant for Disabled in Intelligent
Urban Environment'' Proceeding of IEEE/RSJ International
Conference on Intelligent Robots and Systems, Ballyfs
Las Vegas Hotel, USA, October 27-31, 2003
- 16
- Peter T. Szemes, and Hideki Hashimoto, ``Human Machine
Interface for Intelligent Space'' Proceeding of The 21st
Annual Conference of the Robotics Society of Japan, Tokyo,
Japan, Sept. 20-22, 2003
- 17
- Hideki Hashimoto, Peter T. Szemes, ``Ubiquitous Haptic
Interface in Intelligent Space'' Proceeding of Annual
Conference of The Society of Instrument and Control Engineers,
Aug. 4-6, 2003, Fukui University, Fukui, Japan
- 18
- Peter T. Szemes, Joo-Ho Lee, Hideki Hashimoto, and
Peter Korondi, ``Guiding and Communication Assistant for
Disabled in Intelligent Urban Environment'' Proceeding
of IEEE/ASME International Conference on Advanced Intelligent
Mechatronics, July 20-24, 2003, International Conference
Center, Port Island, Kobe Japan p.598-603
- 19
- Peter T. Szemes, Joo-Ho Lee, Noriaki Ando, and Hideki
Hashimoto, ``Ubiquitous Haptic Interfaces in Intelligent
Space'' Proceeding of The Eight International Symposium
of Artificial Life and Robotics (AROB8th '03) Beppu, Oita,
Japan, 24-26 January, 2003
Micro-Telemanipulation
- 20
- Metin Sitti , Hideki Hashimoto, ``Macro to nano Tele-Mnipulation
towards Nonoelectromechanical Systems'', Journal of Robotics
and Mechatronics, Fuji Technology Press Ltd., Vol.12,
3, pp.209-217, 2000
- 21
- Noriaki Ando, Masahiro Ohta, Hideki Hashimoto, ``Micro
Teleoperation with Haptic Interface'', Proceedings of
the 2000 IEEE International Congerence on Industrial Electronics,
Control and Instrumentation, pp.13-18, 2000
- 22
- Noriaki Ando, Masahiro Ohta, Hideki Hashimoto, ``Micro
Teleoperation with Parallel Maniulator'', Proceedings
of the 2000 IEEE/RSJ International Conference on Intelligent
Robotics and Systems (IROS2000), Vol.1, pp.677-682, 2000
- 23
-
Peter Korondi, Peter T. Szemes, Hideki Hashimoto, ``Sliding
Mode Friction Compensation for a 20 DOF Sensor Glove'',
Journal of Dynamic Systems Measurement and Control, ASME,
Vol.122, 4, pp.611-631, 2000
- 24
- Noriaki Ando, Peter T. Szemes, Peter Korondi, Hashimoto
Hashimoto, ``Improvement of Response Isotropy of Haptic
Interface for Tele-micromanipulation Systems'', Proceedings
of the 2002 IEEE International Conference on Robotics
and Automation, pp.1925-1930, 2002.05, Washington, DC,
ISBN 0-7803-7272-7
- 25
- Peter T. Szemes, Peter Korondi, Noriaki Ando, Hideki
Hashimoto, ``Friction Compensation for Micro Tele-Operation
Systems'', Automatica, Journal of Control, Measurement,
Electronics, Computing and Communications, Vol.42, No.1-2,
pp.23-27, 2001, ISSN 0005-1144
- 26
- Peter T. Szemes, Peter Korondi, Noriaki Ando, Hideki
Hashimoto, ``Master Device For Micro Tele-Operation Systems'',
Proceedings of the 10th IEEE International Conference
on Advanced Robotics (ICAR 2001), pp.369-374, 2001.08,
Budapest, Hungary
- 27
- Peter KORONDI, Noriaki ANDO, Peter T. SZEMES , Hideki
Hashimoto, ``MASTER DEVICE FOR MICRO TELE-OPERATION SYSTEMS'',
Journal of Electrical Engineering, Vol.1, No. 2, pp.57-62,
2001.04, ISSN 1582-4594
Intelligent Transportation Systems Project
- 28
- Massaki Wada, Mami Mizutani, Masaki Saito, Xuchu Mao,
and Hideki Hashimoto, ``iCAN: Pursuing Technology for
Near Future ITS'', IEEE Intelligent System Magazine.(invited
Paper) International Conference Papers
- 29
- Xuchu Mao, Massaki Wada, Hideki Hashimoto, ``Investigation
on Nonlinear Filtering Algorithms for GPS'', IEEE Intelligent
Vehicle Symposium (IVf2002), pp.64-70, 2002.6
- 30
- Xuchu Mao, Massaki Wada, Hideki Hashimoto, ``Nonlinear
Filtering Algorithm for GPS Using Pseudorange and Doppler
Shift Measurements'', The IEEE 5th International Conference
on Intelligent Transportation Systems (ITSCf02),
pp.914-919, 2002,9
- 31
- Xuchu Mao, Massaki Wada, hideki Hashimoto, ``Investigation
on Nonlinear Models for GPS Algorithms'', 9th World Congress
on Intelligent Transport Systems, pp. TP088-3167, 2002.10
- 32
- Xuchu Mao, Massaki Wada, hideki Hashimoto, ``Nonlinear
GPS Models for Position Estimate Using Low-cost Receiver'',
The IEEE 6th International Conference on Intelligent Transportation
Systems (ITSCf03), 2003.10
- 33
- Xuchu Mao, Massaki Wada, hideki Hashimoto, ``EM/Unscented
Smoothing Based Parameter Learning for Nonlinear Models
for GPS Positioning'' 10th World Congress on Intelligent
Transport Systems, 2003.11
Peter Tamas Szemes 2003-09-03
|
|
![\includegraphics[width=4in]{ispace-e.eps}](../../../announcements/Newsletter/JulySep2003/img1.png)
![\includegraphics[width=4in]{dind-e.eps}](../../../announcements/Newsletter/JulySep2003/img2.png)
![\includegraphics[width=4in]{dind-e2.eps}](../../../announcements/Newsletter/JulySep2003/img3.png)
![\includegraphics[width=4in]{pose_estimation.eps}](../../../announcements/Newsletter/JulySep2003/img4.png)
![\includegraphics[width=4in]{color_bar_codes.eps}](../../../announcements/Newsletter/JulySep2003/img5.jpg)
![\includegraphics[width=4in]{Assistant1.eps}](../../../announcements/Newsletter/JulySep2003/img6.png)
![\includegraphics[width=3in]{VRoom.eps}](../../../announcements/Newsletter/JulySep2003/img7.png)
![\includegraphics[width=3in]{pseudorange01.eps}](../../../announcements/Newsletter/JulySep2003/img8.png)
![\includegraphics[width=3in]{doppler02.eps}](../../../announcements/Newsletter/JulySep2003/img9.png)
![\includegraphics[width=4in]{noise_18.eps}](../../../announcements/Newsletter/JulySep2003/img10.png)
![\includegraphics[width=4in]{receiver003.eps}](../../../announcements/Newsletter/JulySep2003/img11.png)
![\includegraphics[width=4in]{concept.eps}](../../../announcements/Newsletter/JulySep2003/img12.png)
![\includegraphics[width=2.5in]{master2.tif.ps}](../../../announcements/Newsletter/JulySep2003/img15.png)
![\includegraphics[width=2.5in]{master_kine.eps}](../../../announcements/Newsletter/JulySep2003/img16.jpg)
![\includegraphics[width=3in]{mrac_fig.eps}](../../../announcements/Newsletter/JulySep2003/img14.jpg)