Home Projects

Projects and Experiences


Milimeter Wave Radar for Robot Navigation in Smoke

we propose the use of single-chip millimeter-wave (mmWave) radar, which is lightweight and inexpensive, for learning-based autonomous navigation. However, because mmWave radar signals are often noisy and sparse, we propose a cross-modal contrastive learning for representation (CM-CLR) method that maximizes the agreement between mmWave radar data and LiDAR data in the training stage to enable autonomous navigation using radar signal.

In the team, I am responsibile for the hardware system of the UGV and execute the experiment. Besides, I also takeover the future work for reconstructing lidar points to mmwave since Gazebo doesn't provide radar informations.

Paper Website


DARPA SubT Urban Challenge

The DARPA Subterranean (SubT) Challenge aims to develop innovative technologies that would augment operations underground. The SubT Challenge will explore new approaches to rapidly map, navigate, search, and exploit complex underground environments. I participated the Urban challenge with my team and our robots in a decommissioned nuclear power plant.

In the team, I am responsible for the UGV hardware system, and the spherical nodes and miniature cars for communication systems.

Paper Video


Deep Reinforcement Learning in Simulation

  • Implementation of different reinforcement learning algorithms such as RDPG and D4PG by PyTorch, the DRL agents were trained by interacting with simulation environment via Gazebo simulator.
  • Using X1 in simulation, and Husky for real robot experiment
  • Obtain navigation skills by training in virtual SubT cave environment
  • I design a narrow gate scenario in gazebo for training in order to urge UGV to pass through narrow passage, and evaluate the performance via sim-to-real approach.


    • Pyrobot-Pick and Place Mission in Simulation(2021/07)

  • Identify the object and pick it up, then move to the target location and place the item.
  • Task 1: Object Detection Mask R-CNN
  • Task 2: Pose Estimation and Pick Dope
  • Task 3: Move to destination A*
  • Task 4: Place in the box

    • I am responsibile for task2 and task3, using Dope for pose estimation, and A* algorithm for goal navigation.


    Report


    MOOS & ROS waypoints navigation

    I audit Marine Autonomy, Sensing and Communications(MOOS-IvP) course given by MIT, the course focus mainly on software and algorithms for autonomous decision making by vehicles operating in the ocean environments. We accomplish the final project of the course at Bamboo Lake, Taiwan. Executing multi-vehicles waypoints navigation by MOOS using Duckieboats which are developed by our laboratory.

    I am responsible for the communication between multi-vehicles and base station.

    Course Website