Page 87 - profile2012.indd

Research Faculty 研究人員

研究員 副研究員
劉庭祿 Tyng-Luh Liu 劉進興 Jing-Sin Liu

Research Fellow Associate Research Fellow
Ph.D., Computer Science, New York University Ph.D., Electrical Engineering, National Taiwan University

Tel: +886-2-2788-3799 ext. 1508 Fax: +886-2-2782-4814 Tel: +886-2-2788-3799 ext. 1813 Fax: +886-2-2782-4814
Email: liutyng@iis.sinica.edu.tw Email: liu@iis.sinica.edu.tw
http://www.iis.sinica.edu.tw/pages/liutyng http://www.iis.sinica.edu.tw/pages/liu

● Research Fellow, IIS, Academia Sinica (2010 - ) ● Research Award for Junior Research Investigator, Academia ● Associate Research Fellow, IIS, Academia Sinica (1994-present) ● MS, Mechanical Engineering, National Taiwan University,1986
● Associate Research Fellow, IIS, Academia Sinica (2005 – 2010) Sinica (2006) ● Assistant Research Fellow, IIS, Academia Sinica (1990-1994) ● BS, Mechanical Engineering, National Cheng-Kung Univer-
● Assistant Research Fellow, IIS, Academia Sinica (1998 – 2005) ● Managing Editor, Journal of Information Science and Engineer- ● Ph.D, Electrical Engineering, National Taiwan University, 1990 sity,1984
● Ph.D. Computer Science, New York University (1997) ing (2010 - )

Research Description Publications Research Description Publications

My research has focused on computer vision and pattern 1. Kai-Yueh Chang, Tyng-Luh Liu, Hwann-Tzong Chen, and We face the challenges of making robots as both a good 1. Ju M. Y., Liu J. S., Shiang S. P., Chien Y. R., Hwang K. S., and

recognition. Speci cally, I am most interested in the topics Shang-Hong Lai. “Fusing Generic Objectness and Visual Sali- partner and a good helper to human daily life in the near Lee W. C., 2001,“Fast and accurate collision detection based
th
of object detection and recognition, image segmentation, ency for Salient Object Detection,” Appeared in the 13 Inter- future, in addition to a key component of industrial auto- on enclosed ellipsoid,”Robotica, vol.19, pp.381-394,2001.
and scene understanding. To tackle the high complexity national Conference on Computer Vision, Barcelona, Spain, mation. To achieve autonomy with desired behaviors, a lot 2. Lin WS, Tsai CH and Liu JS, 2001, “Robust neuro-fuzzy con-
November 2011. (ICCV-2011)

and to design e cient algorithms for practical applica- of issues including sensing, planning and control are go- trol of multivariable ystems by tuning consequent member-

tions, my approaches toward solving the aforementioned 2. Yen-Yu Lin, Tyng-Luh Liu, and Chiou-Shann Fuh. “Multiple ing on to be resolved to a satisfactory level ensuring the ship functions,” Fuzzy Sets and Systems, vol.124, pp.181-195,
vision problems rely extensively on the use of machine Kernel Learning for Dimensionality Reduction,” Appeared in safety and comfort of human-robot interaction and meet- 2001.
IEEE Transactions on Pattern Analysis and Machine Intelli-
learning techniques. gence, vol. 33, no. 6, pp. 1147-1160, 2011. (TPAMI) ing the needs of human beings. 3. Lai HC, Liu JS, Lee DT, Wang LS, 2003, “Design parameters
study on the stability and perception of riding comfort of the

In our recent research e orts, my students and I have been 3. Yen-Yu Lin, Tyng-Luh Liu, and Chiou-Shann Fuh. “Dimen- Along the line of robotics research, at present we conduct electrical motorcycles under rider leaning,” Mechatronics,
working on exploring the visual saliency information for sionality Reduction for Data in Multiple Feature Representa- both simulation and experimental research in the follow- vol.13, pp.49-76, 2003.
more satisfactorily addressing a number of vision tasks. tions,” Advances in Neural Information Processing Systems ing topics: 4. Liang TC, Liu JS, Hung G.-T. and Chang Y.-Z., 2005, “Practi-

The rst problem we investigate is the image co-segmen- 21, edited by D. Koller, Y. Bengio, D. Schuurmans, L. Bot- 1. Control and Optimization: (act while optimizing) to cal and flexible path planning for car-like mobile robot using

tation. By constructing a prior term in the energy function tou, and A. Culotta, pp. 961-968, MIT Press, Cambridge MA, carry out the task such as pursuit-evasion. maximal-curvature cubic spirals,” Robotics and Autonomous
2009. (NIPS-2008)
to account for the co-saliency over a set of images and a Systems, vol.52, no.4, pp.312-335, 2005.
global tem to respect the submodular regularity, we have 4. Yen-Yu Lin, Tyng-Luh Liu, and Chiou-Shann Fuh. “Local 2. Smooth path planning: develop soft computing based 5. Pan WH, Liu JS, Ku WY, 2007, “Fast collision detection for the

established a new and e cient formulation of energy Ensemble Kernel Learning for Object Category Recognition,” (e.g. parallel genetic algorithm based) approaches to scaled convex polyhedral objects with relative motion,”IEEE
Appeared in IEEE Computer Society International Confer-
minimization that leads to state-of-the-art performances ence on Computer Vision and Pattern Recognition, Minneapo- account for physical characteristics of mobile robot Int. Symposium on Assembly and Manufacturing, Ann Arbor,
for image co-segmentation. Building on this success, we lis, MN, USA, June 2007. (CVPR-2007, Oral) motion. Michigan, July, 2007.
are motivated to consider the problem of salient object 5. Hwann-Tzong Chen, Tyng-Luh Liu, and Chiou-Shann Fuh. 3. 3D SLAM (simultaneous localization and map build- 6. Wei JH, Liu JS, 2009, “Mobile Robot Path Planning with eta3
detection. Conceptually, our method can be best under- “Tone Reproduction: A Perspective from Luminance-Driven ing) in indoor environment and its applications in mo- splines Using Spatial-Fitness-Sharing Variable-length Genetic
stood through an integration of two graphical models that Perceptual Grouping,” International Journal of Computer Vi- bile robot navigation. Algorithm,” 2009 IEEE International Conference on Intel-
ligent Robots and Systems, St. Louis, Missouri, USA, Oct.,
correspond to explaining the objectness and saliency esti- sion, vol. 65, no. 1-2, pp. 73-96, November 2005. (IJCV) 2009.
mations, respectively. Via carrying out a two-step optimi- 6. Yen-Yu Lin, Tyng-Luh Liu, and Hwann-Tzong Chen. “Se- 4. Robot-assisted wireless sensor network: data collec-
zation procedure, the proposed technique can iteratively mantic Manifold Learning for Image Retrieval,” Proceedings tion navigation and path planning, localization. 7. Ho YJ, Liu JS, 2010,” Simulated annealing based algorithm
for smooth robot path planning with different kinematic con-
th
improve the quality of the estimations of objectness and of the 13 Annual ACM International Conference on Multi- straints,” 25 ACM Symposium on Applied Computing, Sierre,
th
saliency for a scene, and yields the results of salient object media, pp. 249-258, Singapore, November 2005 (ACM MM- Switzerland, March, 2010.
detection. Our ongoing research along this line is to use 2005, Best Student Papers Session) 8. YS Chou and JS Liu 2011, Indoor 3D Map Building by Laser
the salience cue to improve solving the task of object rec- 7. Hwann-Tzong Chen, Huang-Wei Chang, and Tyng-Luh Liu. Range Finder Using a Four-bar Linkage Rotating Motion Plat-
ognition. “Local Discriminant Embedding and Its Variants,” Appeared form, 11 International Conference on Automation Technol-
th
in IEEE Computer Society International Conference on Com- ogy, Daoliu, Yunlin, Taiwan, Nov. 2011.
The other main research activities in my lab are to develop puter Vision and Pattern Recognition, vol. 2, pp. 846-853, San

speci c computer vision techniques and exploit new vi- Diego, CA, USA, June 2005. (CVPR-2005, Oral) 9. KM Chiu and JS Liu 2012, Path planning of a data mule for
sion applications in all aspects for the next-generation im- 8. Tyng-Luh Liu and Hwann-Tzong Chen. “Real-Time Tracking data collection in the sensor network by using an improved
clustering-based genetic algorithm, 2012 International Con-
aging/video devices such as Kinect and 3-D cameras. Using Trust-Region Methods,” IEEE Transactions on Pattern ference on Affective Computing and Intelligent Interaction
Analysis and Machine Intelligence, vol. 26, no. 3, pp. 397-
(ICACII 2012), Taipei, Taiwan, Feb. 2012.
402, March 2004. (TPAMI)
10. YC Lin, JS Liu, KM Chiu, 2012, A novel hybrid localization
system combining a hexagon-based algorithm and mobile
anchor,” IET International Conference on Automatic Control
and Artifi cial Intelligence, Xiamen, China, Mar. 2012
研究人員
86 Research Faculty
87