Starting from 2018 I worked as an undergraduate research assistant in the HKUST Robotics Institute through UROP, the Undergraduate Research Opportunities Program.

I mostly worked in Professor Shaojie Shen's group, the HKUST Aerial Robotics Group and helped with documentation and tool chain construction and testing, I also wrote several software modules for the Mocka Swarm Project, one of which is displayed here.

This optimization optimizes the map by finding and deleting internal points (those surrounded by other points in all directions) to reduce the total number of points in the PointCloud.

The size of the PointCloud will thus be reduced, which leads to better run-time performance as less data processing are needed.

A cross section of the optimized map is shown on the left, which indicates desired internal voids.

This module takes parameters such as random seed, path width, wall width, and map size with which it will then use the recursive division algorithm to randomly generate a 2D Maze in PointCloud form.

The map can then be used to test the Path Planning Module.

This module takes parameters such as random seed, path width, wall width, cell count,and map size. It will then use a self designed algorithm to randomly generate a 3D Maze in PointCloud form.

The algorithm first creates random cores in the map space, then construct middle planes between these cores to generate a Voronoi Pattern. Finally, passages are cleared between random cells to create a maze.

The map can then be used to test the Path Planning Module.