Fall 2022 Progress Updates

A general summary of all the progress that was made in the Fall of 2022.

Pre-flight Planning

The idea was to generate a occupancy grid of campus that can be used as an occupancy grid.

  • Used OpenStreetMap to generate a isolated 3d map of campus.
  • Converted that into a 2D map using Rasterization techniques.
  • Initially planned on using Octrees, but later decided not to due to the vast amount of memory that would consume.
  • Has a separate page, which describes the stack in much greater detail. /wiki/active-projects/drone-delivery/images/ The following is the point cloud generated from OpenStreetMap. cloud

The Octree that was generated. tree

The occupancy gird that was generated. occupancy

Obstacle Avoidance:

  • Migrate the code for Realsense camera from Python → C++
    • Able to get depth matrix
    • Can convert to occupancy matrix
  • Implimenting a the D* and A* algorithms.
    • There are issues in how the paths are generated, in that sometimes diagonal paths are taken over straight ones, even though the latter is possible to produce.
    • To be improved in Spring 2023.
  • OpenCV integration for detecting obstacles at a greater depth.

The Output from the A* algorithm we develped: astar

An example of an unideal path: erros



  • Repaired the old drone, and interfaced a computer with it.
  • Simulated pre-programmed fight paths using Gazebo and QGroundControl.
  • Worked on importing OpenStreetMap data into Gazebo.



  • Project Managers: Sooraj Chetput
  • Obstacle Avoidance Team Members: Guna Avula (Lead), Deegan Osmundson, Chris Qiu, Ethan Baird, Mouli Sangita
  • Pre-flight planning: Seth Deegan (Lead), Vincent Wang
  • Research: Sreevickrant Sreekanth (Lead), Vignesh Charapalli
  • Hardware: Jacob Harrelson (Lead), Evan Zher
  • Interfacing: Sooraj Chetput (Lead), Atharva Bhide
Last modified February 15, 2023: fixing images (f597d17)