Autonomous Drone Racing
This lab maybe is super cool https://hiperlab.berkeley.edu/research-projects/motion-planning/
To get excited , look at this:
Just watch this video: https://www.youtube.com/watch?v=0kvWwo9StqY&ab_channel=MadhurBehl
- This guy came up with a more recent talk, https://www.youtube.com/watch?v=jv4f_n5D2oQ&t=2569s&ab_channel=ArtificialandMechanicalIntelligence
- This is a shorter video https://www.youtube.com/watch?v=Hebpmadjqn8&ab_channel=UZHRoboticsandPerceptionGroup
https://rpg.ifi.uzh.ch/research_drone_racing.html
The trajectory is generated with an MPC controller.
Links
- https://github.com/mit-aera/Blackbird-Dataset (dataset)
- https://rpg.ifi.uzh.ch/docs/RAL2023_Cioffi.pdf (paper that came out, where they learn the odometry in 2023)
There are actually a couple of courses on this:
- MIT16.485 - Visual Navigation for Autonomous Vehicles
- https://vnav.mit.edu/
- This was the course proposal https://arxiv.org/pdf/2206.00777.pdf
I’ve always been a fan of FPV Drone racing. Maybe I do this project and get an internship at Skydio.
It’s the same challenges. Reduce latency, and there is the Sim2Real problem.
Time-Optimal Trajectory Planning for Quadrotors
look into Flightmare Quadrator simulator (from the madhur behl video)
Some parts that Preston recommended me
- https://docs.px4.io/main/en/flight_controller/cubepilot_cube_orange.html
- https://docs.cubepilot.org/user-guides/autopilot/the-cube-module-overview
Simulations:
https://www.youtube.com/watch?v=p_Q6V0M-Ono&ab_channel=COMBATAPPROVED
From the video
Wowww, they are able to solve the Perception-Aware equations at 100hz using