I recently completed my PhD as a member of the Adaptive Robotic Manipulation Lab at Georgia Tech. I enjoy working on robot planning & control, mechatronic system integration, and automation in general.

Projects

Below are a few projects I’ve worked on as part of my graduate research, for coursework, or just for fun.

Jumping robot trajectory optimization

In this project, we planned jumps for a pneumatic muscle–driven legged robot. We implemented direct collocation with hybrid robot dynamics (robot feet on vs. off ground) using the GTSAM factor graph library. We used a soft model to approximate the hybrid pressure dynamics (valves open vs. closed), in which time-shifted sigmoids ‘activate’ or ‘deactivate’ each set of pressure dynamics based on valve state.

Optical sensor design for pneumatic muscle state estimation

I developed a ‘pneumatic artificial muscle’ actuator with integrated optical sensors for estimation of muscle contraction and force. Each optical sensor uses an LED-photodiode pair to measure the light reflected by a silicone diaphragm embedded in the muscle.

Particle filter for 2D robot localization

I implemented a particle filter for localizing a robot inside a building—a map of the building is given, but there is no prior knowledge of robot pose. I develped motion and sensor models for robot odometry and LIDAR data, used ray-casting to determine true range measurements, and performed importance sampling to resample particles at each timestep.

Balancing wheel

For this project, affectionately called the ‘hamster wheel,’ we built a two-wheeled system in which a free wheel is balanced on top of a motor-driven base wheel. The system is linearized around the free wheel’s top position and an LQR controller is implemented.

Balancing robot

I built a simple balancing robot driven by two DC gearmotors with encoders. An IMU was used to estimate robot orientation via a complementary filter. Similar to the balancing wheel above, an LQR controller was implemented on an MSP432 microcontroller.