DROVER: Drone-Rover Communication for Pathfinding
Noah Bergam, Justin Li, Micah Elwyn, Sam Henriques, Dr. Jolly
Project Summary
Due to their unique abilities and visual perspectives, integrating a drone and a rover into a single autonomous system is an enticing prospect, with applications currently being explored in wildfire tracking, garbage collection, and extraterrestrial reconnaissance
[1], [2], [3]. This project, entitled DROVER, is concerned with building such a system cheaply and effectively and applying the AirGround protocol for drone-rover communication, developed by [4], to a pathfinding problem in an unknown environment. The main challenge in solving this problem is implementing a multi-robot simultaneous localization and mapping (SLAM) protocol which can integrate the very different perspectives of the drone and the rover in real time. Once physically active, the DROVER system will be tested based on its speed of various path traversals with respect to different multi-robot SLAM algorithms, such as Extended Kalman Filters (EKF), Extended Information Filters, and Particle Filters (PF) [5].
Figure 1. Drone setup
Figure 2. Rover
Figure 3. Stereo system
Methods and Direction
Most of the work put into the project thus far has been concerned with physical design of the system and data intake. We currently have a drone and a rover which we plan to network using Raspberry Pi (see pictures). We are currently working on calibrating a stereo-vision system for the rover and we have done initial testing on pathfinding algorithms for the drone (see Test1 and Test2 folders).
With the physical part of the project interrupted due to the quarantine, our focus has more recently been directed towards algorithm development. We plan to find or create footage to use for SLAM algorithm testing, and develop the backbone for the code which would be loaded on the drone and the rover, integrating ideas from our previous coding work. We may use online robotics simulations for testing.
Drover Test 1: Absolute Image Tracking.
Drover Test 2: Weighted Image Tracing
References
[1]: https://www.colorado.edu/aerospace/sites/default/files/attached-files/drift_msr.pdf
[2]:https://www.theverge.com/2016/2/25/11112160/volvo-robot-drone-trash-prototype
[3]: https://www.vox.com/recode/2020/7/30/21348560/mars-rover-nasa-perseverance-autonomous-helicopter-drone
[4]: https://ieeexplore.ieee.org/document/7230963
[5]: https://onlinelibrary.wiley.com/doi/abs/10.1002/rob.21620
Meet the Team