Carnegie Mellon Robotics Institute
I'm interested in dynamics and control problems in robotics and manufacturing automation. Here are some aspects of my recent research.
Space robotics: Robotic technology is potentially beneficial for safer, more productive, and more economical exploration in space. We have developed the Self-Mobile Space Manipulator (SM2) which was an autonomous walking robot that is capable of both locomotion and manipulation. The telerobotic testbed we developed include a zero gravity environment, three prototypes of robots, and a ground-base teleoperation control station. Lately we have been working on extension of SM^2 for lunar operations, i.e., DM2 (the Dual-use Mobile Detachable Manipulator) which can detach from a mobile base and climb on lunar structures, or connect to other mobile base to form a train-like configuration for both exploration and maintenance tasks in a lunar base. The research issues in this area that I'm interested are: (a) dynamic coupling of the robot systems and space vehicles, the use of the dynamic coupling in controlling the attitude of vehicles and robot motion, (b) the design and control of space robots for both locomotion and manipulation, i.e., the robot with self mobility and capability of performing simple tasks, and (c) human interfacing and telerobotics, and real-time shared control of autonomous control and teleoperation.
Control: My interest in control lies on the intelligent control that can learn in real-time in unconstructed environment. More specifically, we are working on learning and transferring human control strategies in response to real-time inputs. We are investigating the following issues: (1) how to efficiently model human control strategy, (2) how to select state inputs in order to generate reliable models, (3) how to validate the computed models, (4) how to evaluate the quality of the HCS models, (5) how to optimize performance of the HCS models, and (6) how to effectively transfer human control strategy to a robot or a human. Learning can be done in a virtual environment and skill can be transferred to a real operation of a robot system. The research on this line has been increasingly recognized and will lead to tremendous applications in intelligent highway/vehicle systems, human-machine interface, realtime training, virtual reality, and game industries.
Mechatronics: I have been interested in integration of sensing, actuation, control, and machine intelligence for high-performance machine systems for a long time. My past experience in developing compliant wrist and CMU rapid assembly system allowed me to approach problems in systematic way to integrate the design, planning, low-level control, and real-time control. More recently, we have been working on the robust sensing and control of planar linear motors and investigating the application in agile manufacturing. We are also developing the technology for visual controlled welding robot system with real-time learning capability.
Dynamics: The dynamically stable, but statically unstable systems could be useful in high-speed, rough terrain, and tough environment. We developed a single-wheel robot using the gyroscopically stabilization principle. We are also developing the control schemes for underactuated robots with both passive and active joints utilizing dynamic coupling of each joint. Control schemes and mechanical hardware to alleviate the nonholonomic problems associated to these systems is interesting to me. The research on free-floating robots in zero-gravity environment can also be considered as an example of dynamically stable robot. The dynamic nature of the system is interesting and its utilization in robotics has not been extensively explored.
|Research Interest Keywords|
|control, design, factory and warehouse automation, space robotics, teleoperation|
|The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.|
Contact Us | Update Instructions