Research PrograM

My research program explores the cyclic learning cycle between biology, mathematical system theory, and robotics with an emphasis on control theoretic aspects of bio-inspired robots with extreme behaviors

Control of Bio-Inspired Robotics

static_openloop_8 (1).mp4

Generalized Flapping-Wing Vehicle Control

The goal of this project is to develop a generalizable safety-critical adaptive control framework for systems with extreme behaviors capable of leveraging the environment through strategic interactions. This project will investigate the control of a large class of flapping wing vehicles with scalable numbers of wings and with multiple actuation methods, such as controlling the abdomen, to research energy-efficient maneuver strategies. In particular, the mixture of low-level and high-level controllers takes advantage of surrounding air flow to change the control policy via optimization techniques adaptively and demonstrate agile maneuvering in the natural environment. Moving forward, the goal is to investigate the social behavior of flapping-wing vehicles by constructing a new signaling method (e.g., the pose of an individual or swarm as a message) among several teams of swarm FWVs. 

Impulsive Systems Control

Many insects use latches as a mediating tool to rapidly release stored potential energy achieving remarkable impulsive behaviors (e.g., mantis shrimp strikes, grasshopper jumps).  This project will further investigate the LaMSA (Latch Mediated Spring Actuation) framework that describes energy transfer from the slowly loaded spring to an impulsive change in kinetic energy, which is an essential power amplification strategy for small-scale organisms. My recent work on the generalization of controllable behavior in ultra-fast systems explores numerous directions of tunability in LaMSA systems, affecting the output performance. This research allows us to further investigate the optimal design of LaMSA mechanisms and the generalized role of feedback control as impulsive motion occurs in milliseconds. I envision that the study of feedback control of these impulsive systems will provide a fundamental framework for controlling impulsive systems accompanied by challenges for high-speed perception and computation. 

Swarm robotics with flapping-wing vehicles


The enhanced safety-critical controllability of flapping-wing vehicles (FWVs) provides us with a framework to apply to the swarm of FWVs, resembling the collective behavior of biological insects. Inspired by the collective behavior of bird flocking and the communication in the waggle dances of bees, this project would investigate distributed swam control studying the categoric primitives of geometric formation as a potential tool for relaying information. The first challenge in this study is to design a decentralized formation controller for highly dynamic systems with limited sensing, power, and computation. This research will further investigate the class of formations which can be more robust under disturbance such as surrounding airflow, and learn from the distorted formation to estimate the airflow. This information encryption through group behavior will not be limited to FWVs but can also define coordination within heterogeneous group of robots such as ground robots.  

Impulsive System Theory

Extreme (e.g., impulsive) behaviors in robotics can also occur from interactions with rigid environmental objects.  The system dynamics with s in state (i.e., changes in velocities before and after the foot contact with the ground for bipedal walking robots) are often modeled by as a hybrid system which preassumes the effect of impulsive collisions rather than causally modeling the impulsive contact forces within the dynamics. However, defining the impulsive contact force with singular functions (e.g., Dirac delta function) is challenging since the singular functions have limited usage for nonlinear systems (Schwartz impossibility theorem ). My past research proposed a new generalized function framework base on a branch in mathematics called non-standard analysis to rigorously define the impulsive function (a model for the impulsive contact force) to represent a causal model for nonlinear systems with discrete changes. This new framework emphasizes the feasibility of impulsive motions by describing the contact forces explicitly in mechanical modeling with extreme behaviors.