My research interests are generally on robotics and control. Currently, I am doing research on bio-inspired jumping robots, multi-mode locomotion for miniature robots, and visual servoing using non-vector space control theory. Please refer to the following for the projects I have participated in. You may also want to browse my publications.

MSU TailBot: A Tail Assisted Running and Jumping Robot

This research aims to investigate the multi-modal locomotion ability for small robots. Specifically, we want to design a robot that can run on flat ground, jump to overcome large obstacles, and control its mid-air orientation once leaping into the air.

We have built a miniature robot which can perform the above functions. It has a lightweight 26.5 gram and a small maximum size 7.5 cm, yet it can achieve the three functions: run, jump, and aerial maneuvering. Furthermore, this robot is equipped with on-board energy, sensing, control, and wireless communication capabilities, which enables the tetherless operation. Read More

MSU Jumper: A Single-motor-actuated Miniature Jumping Robot

The goal of this project is to design miniature robots to mimic the jumping abilities widely found in small animals such as fleas, grasshoppers, and frogs. With jumping, the robot can overcome large obstacles compared to its small size. We have built three generations of jumping robot. The latest generation is shown on the right.

The robot can repeatedly perform the following motion sequence: First of all, it can rotate on the ground to a desired jumping direction. After that, it can charge energy to springs and self-right from the landing posture simultaneously. Then the robot jumps into the air by instantly releasing the energy stored in the spring. Finally, the robot lands on the ground, and the next motion cycle starts. Read More

Visual Servoing Using Non-vector Space Control Theory

The motivation of this project comes from finding visual servoing methods that are free of feature extraction. For traditional image based visual servoing methods, prominent features are first extracted from the image, and then a controller is designed to make the vector of feature positions converge to a desired value.

Considering the image as sets, we formulate the problem in the non-vector space and design stabilizing controller in this space. The general structure of this approach is shown on the right figure. Read More