A Dual-Objective NMPC Scheme

A sequential dual-objective NMPC scheme applied to a planar quadrotor. Optimize time-optimal motion, stabilizing a target neighborhood of a state-space manifold.

Requirements

• A Linux OS
• matlab
• python (>=3), with numpy and matplotlib
• swig, tested with v3.0.12
• gcc, g++
• blas
• lapack
• casadi, tested with v3.3.0-rc2
• acados, tested with bd193f3
  • blasfeo, tested with 4815341
  • hpipm, tested with 2eb8b1f

How to use

• From Matlab, execute code_generation/generate_code.m
• In a terminal, browse to controller_library
  • Point to python include directory: export PYTHONINC=..., e.g.export PYTHONINC=/usr/include/python3.6m
  • Point to acados installation directory: export ACADOS=..., i.e. where the include and lib directories live (such as the root dir of the GIT repo after compilation).
  • and execute make install
• Use Python to run simulation/simulate.py for simulations

Remarks

• Note that currently only Nta < Ntr is supported, i.e. the economic prediction horizon must be shorter than the manifold stabilizing prediction horizon.
• The lapack and blas libraries are assumed to be installed in /usr/lib/.