... and other multicopters.
Written as part of my degree project, this software is based on PyQtGraph for 3D live viewing of the simulation, and Numba for high-speed matrix heavy computation. This project includes a PID regulator-based controller inspired by the PX4 flight stack by Meier et al.
It also includes an implementation of the iLQR algorithm, a trajectory optimization method based on DDP. Trajectory optimization is actually what the degree project is about.
Prerequisites: Not Microsoft Windows. Python 3.x.
As this is definitely a package you're going to tinker with, I recommend checking it out and installing a development version like so:
git clone https://github.com/lericson/pysquad
cd pysquad
pip3 install -e .Perhaps slap on a virtualenv if you're that kind of a person.
Live simulation with the PX4-like controller
python3 -m squad.dyn squad.dyn.mc.Quad squad.px4.AgentSame with the DDP-based trajectory optimizer
python3 -m squad.dyn squad.dyn.mc.Quad squad.ddp.make_quad_policyVisualization of trajectory optimization (recommended viewing, very satisfying)
python3 -m squad.dyn squad.dyn.mc.Quad squad.ddp.make_quad_policyQ: There are questionable code practices in this software, and my OCD is itching.
A: Yes, it's a degree project, so focuses on the results and theory rather than code quality. Fix the problems and send a pull request!
Q: You seem to have reinvented argparse. Why?
A: It was easier to implement.
Q: Why is there a FAQ, nobody asked these questions.
A: That's not a question. Perhaps it should be called a PAQ. Probably asked question. Or MLAQ. Maximum likelihood asked question.
Q: Can I read the thesis you wrote on this stuff?
A: Let me just finish it first.
Q: The DDP thingy is too slow.
A: Yes, that was what my thesis was about.
