[Japanese] [Lab. Home] [OESE Home] [CIST Home]

Naoki Oda Chitose Institute of Science and Technology 758-65, Bibi, Chitose, Hokkaido, 066-8655 TEL and FAX 81-123-27-6104

Miscellaneous Tips for Students

♣ OpenGL Programing using MFC ( in Japanese )
♣ Socket Programing(Win,UNIX) ( PDF Japanese )

♣ Primary Domain Controller(PDC) using Samba ( in Japanese )
♣ Socket Programing(Win,UNIX) ( in Japanese )

Research topics

Research topics is Moved...

Biped Walking Robots...

We are developing a biped motion system. For certain stabilty of biped motion, the recognition of floor condition is one of important issues. We are now developing the laser based sensing system for biped robot. Actually we have made a biped robot below which has 6DOF/leg.
More movies: see Odalab Home Page.

Moving Picture: Laser Sensor Based Walking (WMV9) (Rapid)
Moving Picture: Walking Motion (WMV9) (Slow)
Moving Picture: Simulation(WMV)
. .


Specification of Biped Robot :
6 DC Motors per leg( 60W X 6). Gear: Harmonic Drive(2X 1/120 , 4X1/100 per leg)
Upright Height: 0.90m ( from ground to Hip Plate), Weight: approximately 40Kg
CPU: Pentium4 2.8GHz , OS: RT-Linux, Network Boot without HDD, IO: 16ch Encoder Couner/16chDA/16chAD/16chPWM

Mobile Robots with fully remote control

Specification of Robotic WheelChair :
2 DD Motors, Weight: approximately 90Kg
CPU: Pentium D 950 3.4GHz(Windows) and Pentium M 1.7GHz(RTLinux)
OS: Windows for Image Processing , RT-Linux for Motor Controller
Stereo Cameras: High Speed IEEE1394 CCD Camera x2




Specification of Wheeled Robot :
2 AC Motors. Gear: Harmonic Drive(2X 1/50), Weight: approximately 20Kg
CPU: Celeron 800MHz , OS: RT-Linux





Specification of Small Mobile Robot:
CPU: Crusoe 540MHz , OS: Linux ( It is customized RedHat-Linux), Disk: Compact Flash DISK 128MB, Network: 11Mbps Wireless LAN
Sensor: Color CCD Camera( captured by BT878 Card. Video Image can be broadcasted through Network( 10Frame/sec) ), Photo Distance Sensor
Battery: 12V 2600mAh, Motor: 540Series(Mabuchi) armature current is controlled by DA output.
Steering: Steer angle is controlled by DA output.

Sample Videos...

Video is available from intranet ...
.

Motion Control of Redundant Manipulator

This topics involves a robust control method of kinematically redundant manipulator in order to distribute the motion controller into the simple components. In the redundant system, the flexible utilization of the redundancy is important to realize the dexterous motion such as obstacle avoidance, optimal configuration, highly manipulability and so on. Recently the several robust control strategies using disturbance observer have been proposed, and make it possible to realize the angular acceleration control of each actuator. In these method, the disturbance acceleration or torque is estimated and suppressed by its feed-back.
One of the inverse kinematics calculation in the redundant manipulator is mathematically resolved through the least-norm minimization. This means that the generalized inverse matrix and null space vector of Jacobian are used for the transformation from the desired workspace acceleration into the joint angular one. In another approach, the use of the augmented Jacobian matrix is convenient way to deal with the user-defined additional task for the utilization of the redundant motion. And its calculation process becomes simple algorithm in the same way as non-redundant systems. However, in these approach mentioned above, the motion controller of the redundant manipulator sometimes falls into inflexible control law for only specified tasks through the redundancy resolution. That is to say, the flexibility of the task selection which decides how to use the redundancy is not considered in the motion controller. And it causes the difficulty of the planning of the redundant motion, even though there are capability that the redundancy contributes to the dexterity and multi-behavior functions like human arm. This research is focused on such practical complexity and difficulty of the redundant system to implement the autonomous selection of the appropriate user-defined task vector according how to use the redundant degree-of-freedom now. Currentrly we are trying to use Optical Sensing Device for Force Based Motion Control. Following Moving Picture shows a compliant motion control using natural stifness of whole manipulator. That is to say, whole of manipulator works as force sensor by measuring torsional deformations.
Moving Picture: Compliant Motion based on Optical Sensing
Moving Picture: Obstacle Aboidance




Specification of 3Link Redundant Manipulator :
3 AC Motors( Harmonic Drive ), PSD Sensor and 4 LED Target available (Hamamatsu Photonics)
OS: VxWorks 5.4




Specification of 2Link DD(Direct Drive) Manipulator:
3 DD Motors( Yogogawa ), Control OS: VxWorks 5.4

Visual Servoing

The recent development of visual image processing is making possibility to realize the real-time motion tracking by visual servoing. In the visual servoing, several constraints between the mechanical structure of the manipulator and the vision sensor coordinate exist according to the algorithm of the target visual image processing. Furthermore, the tracking performance depends on the image processing speed, accuracy and the kinematic characteristic mapped on the image space. In the practical point of view, the visual motion to get the desired image area obtained by visual sensor is applicable to various kind of robot systems such as manipulators, mobile, walking robots and so on. Therefore, the development of the image processing technique and speed will make remarkable progress in industrial field and brings highly contributions on the robot system. So it is expected to consider and extend the robust motion controller based on the visual image space for the optimized configuration of the motion distribution.

Currently we are using the PSD sensor (Position Sensitive Detector) for real-time processing to develop the motion controller based on image plane. In our approach, the workspace observers is directly applied in vision space. Then the dynamical manipulability is considered on the visual plane. And we are planning to develop the weighting approach of the image processing area related to desired motion.



CG Demo Program( Executable )
(for Windows95/98/NT/2000. Recommendation: Grapic Card supporting Hardware OpenGL API)


Specification of 5DOF Manipulator:
MoveMaster ( Mitsubishi ), 5 degrees-of-freedom, 5 DC Motors, each armature current can be controled by PC.
OS: VxWorks 5.4. Image Processor: Hitachi IP-5010 installed on Linux Computer.

Image Sharing System for Cooperative Manipulation of Robots

For cooperative manipulation, we focus on sharing remote vision sensor. In Robot system, the real sharing can be realized by computer network communication. So we are now developing the original video broadcasting system, and designing for the distributed image processing.

Specification of Video Broadcasting System :
Capture System : BT878 Card on Linux Computer, Driver : Video4Linux Version.2
Compression : JPEG Compression per Frame like Motion JPEG format, Protocol : Original .
Transfer Speed : 30Frame/s ( approximately 1.5Mbps ), Software : originaly developed in Oda Lab.

© Oda Laboratory. ALL Rights Reserved.