Taoran's project gallery
Aerobat: Dynamic Morphing Flight
Firmware development and Visual-Inertial Mapping
Related publication：
Gupta B, Shah Y, Liu T, et al. Banking Turn of High-DOF Dynamic Morphing Wing Flight by Shifting Structure Response Using Optimization[J]. arXiv preprint arXiv:2405.05490, 2024.
Sihite E, Ramezani A. A morphology-centered view towards describing bats dynamically versatile wing conformations. The International Journal of Robotics Research. 2024;0(0). doi:10.1177/02783649241272132
One of my recent projects is the development of the perception functionality for Project Aerobat, a flapping-wing aircraft designed for agile flight. Under the supervision of Professor Alireza Ramezani, I contributed to the design and integration of a lightweight visual-inertial perception unit, combining a global shutter camera, IMU sensor, and trigger-based synchronization. This system provides real-time feedback and control for Aerobat’s navigation and stability.
In addition, I developed a quadcopter-based guard for altitude control, utilizing OptiTrack’s pose data and onboard IMU measurements. By implementing PID control, this guard assists Aerobat in maintaining stable hover and precise positioning.
Furthermore, I programmed and debugged Aerobat’s soft PCB wings, focusing on real-time accelerometer and gyroscope data processing via two-wire debug console. This work was conducted at Northeastern University’s Silicon Synapse Lab, as part of my role as a full-time graduate research assistant.
Harpy: Thruster-assisted Bipedal Robot
Embedded system development and Sensor Fusion
Project Harpy is a thruster-assisted bipedal robot designed for advanced locomotion, including walking and jumping. I developed the software and hardware framework for real-time data transmission, utilizing the STM32 microcontroller, OptiTrack motion capture system, and EtherCAT protocol to ensure low-latency communication and control.
I designed the firmware for the STM32-Nucleo and ESP32 microcontrollers with an EtherCAT shield, enabling efficient data streaming for real-time control. Additionally, I developed a data parser using the NatNet protocol to process OptiTrack motion capture data, which is transmitted via Wi-Fi UDP to the ESP32.
I also contributed to the development of a state estimator for the robot, employing an Extended Kalman Filter (EKF) to estimate the robot’s body pose. This estimator integrates motion capture coordinates with IMU sensor readings, providing accurate pose estimation for enhanced stability and performance during locomotion.
Related Publication:
Pitroda, Shreyansh, et al. "Capture Point Control in Thruster-Assisted Bipedal Locomotion." 2024 IEEE International Conference on Advanced Intelligent Mechatronics (AIM). IEEE, 2024.
“Posture manipulation of thruster-enhanced bipedal robot performing dynamic wall-jumping using model predictive control”, 2024 IEEE-RAS 23rd International Conference on Humanoid Robots (Humanoids).
Husky: Thruster-Assisted Quadruped Robot
Embedded system and software development
In this project, I participated in the development of Husky, a thruster-assisted quadruped robot capable of walking, flying, and navigating narrow paths. My contributions included hardware design, where I helped integrate NVIDIA Jetson and Intel RealSense technology for narrow path detection and data parsing from the OptiTrack system using the EtherCAT protocol. I also assisted in creating a hardware testbed system to characterize ducted propellers and implemented the corresponding software drivers.
On the software side, I contributed to the creation of a gait generator that optimizes foot placement based on narrow path detection, enhancing the robot's ability to traverse complex environments effectively. This combination of hardware and software innovations enables Husky to adaptively navigate a variety of terrains.
Rope-Driven Smart Hand for Motion Imitation
Modeling, Control and System Implementation
As part of my graduate project at the Zhiyuan Lab, Beijing University of Chemical Technology, I designed and prototyped a rope-driven bionic smart hand, along with a data collection glove system. This smart hand was developed to imitate human hand movements using a custom-built hardware system.
The system was modeled using ROS and Gazebo, allowing for advanced simulation of real-world dynamics. I also developed an embedded hardware system to validate the hardware's performance in experimental conditions, confirming the system's effectiveness in motion imitation.
Modular Inchworm-Like Soft Crawling Robot
Modeling, Control and Analysis
In this project, I developed a modular inchworm-like soft robot designed for vertical crawling. The robot features adjustable body segments and limb lengths, enhancing its adaptability to various environments. Its movement is driven by multi-chamber and single-chamber actuators that control both bending and extension. I utilized ABAQUS for Finite Element Analysis (FEA) to optimize the actuators’ bending performance for efficient load-bearing during crawling.
Inspired by inchworm locomotion, the robot alternates between bending and straightening states, allowing it to grasp, climb, and traverse vertical surfaces. I designed both the torso and gripper modules, developed mathematical models to analyze actuator performance, and validated the robot’s capabilities through experimental testing.