Fall 2022 Progress Updates

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.

    • https://www.openstreetmap.org/#map=5/38.007/-95.844

  • 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.

The Octree that was generated.

The occupancy gird that was generated.

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:

An example of an unideal path:

Hardware:

  • Established mission objectives, such as flight time and pay load weight.

  • Picked a drone kit, selected baterries, motor controllers, motors and propellers.

    • A detailed document describing all the design choices that were made: https://docs.google.com/document/d/1YcgpvD2AsxBpSHcqvRHN5PRzlyk0kKi2nl-Srrc2DXE/edit?usp=sharing

  • Check the main page for details on what hardware components were selected.

Interfacing:

  • 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.

Research:

  • Over the course of the semester, the team reseached several important topics and developed whitepapers, and plan documents.

    • Deegan Osmundson: A* based navigation algorithms. https://drive.google.com/file/d/1XcB0w0IvobgjAYehDYUqe3qoPYX0miRA/view

    • Seth Deegan: Drone Delivery Tech Stack https://docs.google.com/document/d/1ekadDu0ogtgF6m-fK_AaudN6HzlPpPMSJ_rsvxW74-M/edit#heading=h.fliy5digh3xk

    • Sooraj Chetput: Steps in implimenting the software stack for DD. https://drive.google.com/file/d/1RyhzLklxlVaUF0ReHZfJ17z1Qm90ic8P/view

    • Jake Harrelson and several Authors: Design and Implementation of Unmanned Aerial Vehicle for Local Food Delivery https://docs.google.com/document/d/1YcgpvD2AsxBpSHcqvRHN5PRzlyk0kKi2nl-Srrc2DXE/view

Acknowledgements:

  • 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

Autonomous Robotics Club, 2024