Flight, FORTRAN, Control and Go
summary
Built a thrust‑vectoring VTOL test rig (EDF on a gimbal, ~140 A at 48 V) for Leia Aerospace and achieves tethered lift and control but suffers at least one wreck from an I2C race that hangs all cores. Control uses LQR computed from linearized A/B matrices and a newly ported continuous algebraic Riccati care solver into Gonum by translating ~6,000 lines of Fortran (DTGSEN/DGGES LAPACK logic), which also uncovers real LAPACK bugs. State estimation runs a hybrid particle + Madgwick filter; TinyGo on MCU is proposed to fix timing.
AI-written summary · about
chapters (18)
- 0:00Introduction and agenda
- 0:42What is a VTOL vehicle
- 2:50Building a system model
- 4:14Rigid body dynamics math
- 7:21Optimal control with LQR
- 11:17State estimation challenges
- 13:01Go firmware in action
- 14:14Debugging I²C issues
- 15:17Switching to TinyGo
- 16:19Unifying simulation and firmware
- 17:44Porting Fortran CARE solver to Go
- 19:07Fortran vs Go comparison
- 20:12Testing philosophy differences
- 21:15Finding bugs in LAPACK
- 21:58Mentoring new contributors
- 22:19Future project milestones
- 22:39Raspberry Pi I²C bug
- 23:42Go community ideals
related talks
56mGo Back To The Future
42mRobotics with Go is a Breeze
22mGoDrone: Writing a Quadcopter firmware in Go
26m96Boards Drone Project
46mPaparazzi, An Open Source UAV for Research
53mEfficient Visual Avoidance and Control for UAVs with Embedded GPUs
45mTutorial: Building a Drone, From Scratch
43mFlying robots, a wonderful obsession