Chenggang Liu

   

PhD student, Research Center of Intelligent Robotics, Shanghai Jiao Tong University

Department of Automation, Shanghai Jiao Tong Univ.,800 Dongchuan Rd, Shanghai, China, 200240
Lab Phone:86-21-34204274

frankliu@sjtu.edu.cn

"I know how to program and how to make a chip, now I want to explore the boundaries of these two and the way to generate a silicon-based intelligent life!"
  The great end of life is not knowledge, but action.
                                              --- Thomas Henry Huxley

Research Interests

Currently I am developing a full-body, 25 DOFs humanoid robot with other members. My responsibilities include control system design and hardware system implementation. Project homepage here. You can also find some photos about this project here.

Current Research

Sensory-motor Coordination

Cognitive processes at high abstraction levels rely on a hierarchy of lower level behaviors. Low level autonomous behaviors can be constructed from basic sensorimotor behaviors. A basic sensorimotor behavior is a reflex, which is a direct motor response to sensory stimuli. Basic behaviors are fundamental to a behavior based robot. If basic behaviors are determined a priori by robot designers, the final controller is inefficient at best, or completely wrong at worst. An alternative to overcome these limitations is to acquire basic behaviors by the robot itself.

Because of the experimental accessibility, the plant's simplicity, and the diverse collateral knowledge about the visual system, gaze control is the best-studied control system. Gaze control skills are between instinct and experienced, they provide a promising domain for the investigation of control theory, algorithms and neural mechanisms underlying the acquisition and performance of general sensorimotor skills.

Humanoid Robot Motion Control and Skill Learning

In order to enable humanoid robots to assist us in our daily life, the ability to acquire motor skills will determine its success or failure in the marketplace. Manual programming of humanoid robots for even simple motor tasks in a dynamic environment is an awkward endeavor or even impossible task.

How our central nervous system guides our body? Does our brain carries out similar inverse dynamics calculations? In humans, our mechanical behavior is strongly state-dependent. There may be some building blocks, called motor primitives, in our low level motor control system. The central nervous system may create seemingly complex behaviors by combining a relatively small number of simpler motor primitives. We are trying to find some efficient learning methods in motor control. These approaches are based upon motor primitives and are being applied to motor skill learning in humanoid robotics and legged locomotion.

Research goal

Selected Presentation Slides

Some Photos

Other Interests

My dissatisfaction with the current living styles drive me to create better ones. The following are some programs written by myself. You can download and use them freely. Wish they are useful to you.

  1. Arpm.h and Arpm.cpp implement a C++ class of AR model for the time series analysis. They are compiled with VC++ 6.0 and have been extensively tested under Win2000/XP.
  2. If I am a programmer of Microsoft Co., the first thing I want to do is to update its File Management System (FMS) of Windows OS. One file can only have one name, belong to one directory, have limited information. These shortcomings force me to write this program. It is written with VB6.0 and has been tested for more than one year. The key idea is to save a file with a file ID, and manage the file by a database engine. You can download here (about 23M and only with Chinese help now).
  3. This program is mentioned in the paper 'Basic Behavior Acquisition Based on Multisensor Integration of a Robot Head'. It implements human face detection, the method of sparse iterative version of Lucas-Kanade optical flow in pyramids, and the CAN bus communication. Download here.
  4. Some emacs-lisp functions here.