Dexterous Manipulation (Dynamic Grasping)

1. Designed a general dynamic hybrid manipulation strategy for dynamic grasping.

2. Customized a double inverted pendulum model-based algo for UR16e.

Dexterous Manipulation Online Planning

1. Comprehensive and robust manipulation of objects

2. Fast enough to support online computation and replanning

3. A base for manipulation replanning and recovery

Human-Like Motion Planning Algorithm

1. Analyze human motion and conclude its mechanism i.e coordinate movement and compliant movement

2. Analyze coordinate movement and implement it on 2-d translation and rotation motion

3. Came up with a parametric optimal control algorithm used for this project

Rough Terrain Ground Vehicle Control

1. Implemented algorithms from scratch on CARLA simulation to build a time-aligned dataset, including collecting, transforming and real-time plotting sensor data.

2. Implemented a CNN & LSTM based model from scratch to make motion prediction of an autonomous ground vehicle in urban area and arbitrary rough terrain area. The imported rough terrain map is customized on RoadRunner.

3. Implemented a MPPI-Control based algorithm to guide the motion planning and control of the vehicle.

Optimal Control of Upper Body in Humanoid Robot

1. Developed an analytical bang-bang control solution for a humanoid robotic system that uses variable damping-based impedance coordinative control to optimize lower limb energy consumption caused by upper body oscillation while enabling passive stability of objects held on upper arms, adapting to uncertain lower limb motion.

Fairness-Oriented Control Framework for Safety-Critical Multi-Robot Systems

1. An Alternative Authority Control (AAC) framework is introduced, facilitating the dynamic distribution of control authority among robotic agents.

2. A Model Predictive Control with Flexible Control Barrier Functions (MPC-FCBF) framework is proposed to enhance dynamic obstacle avoidance capabilities.

3. A hierarchical control architecture is developed by integrating the AAC and MPC-FCBF frameworks, offering efficient dynamic authority allocation and improved obstacle avoidance for multi-agent robotic systems.