Skip to content

Conversation

@sedith
Copy link

@sedith sedith commented Mar 19, 2020

Note that these changes have only been tested on the inverted pendulum, using casadi model not exported with the newest version of MATMPC.
Also, this was tested using qpOASES 3.2.1

update cppmpc to be usable with casadi_src file generated by MATMPC@299a4d8

add gitignore file ;
udpate casadi_src.cpp and .h ;
update casadi_wrapper accordingly (Ji->Hi, JN->HN) ;
remove now unused qp_workspace class ;
add qp_problem::info method computing OBJ value (cost function value
among all shooting points) ;
add OBJ double field in rti_step_workspace, also add ::info method ;
considere weights as vector as in MATMPC model generation : WN is a
vector, W is a matrix where the ith collumn is the weight for the ith
shooting point ;
generate control bounds for the nbu_idx (similar to state bounds)
use exported casadi symbols to compute hessians instead of using
intermediate jacobians ;
modify InvertedPendulum.cpp according to all these changes, and add
display of OBJective function in output file

sedith added 2 commits March 19, 2020 19:50
…99a4d8

add gitignore file ;
udpate casadi_src.cpp and .h ;
update casadi_wrapper accordingly (Ji->Hi, JN->HN) ;
remove now unused qp_workspace class ;
add qp_problem::info method computing OBJ value (cost function value
among all shooting points) ;
add OBJ double field in rti_step_workspace, also add ::info method ;
considere weights as vector as in MATMPC model generation : WN is a
vector, W is a matrix where the ith collumn is the weight for the ith
shooting point ;
generate control bounds for the nbu_idx (similar to state bounds)
use exported casadi symbols to compute hessians instead of using
intermediate jacobians ;
modify InvertedPendulum.cpp according to all these changes, and add
display of OBJective function in output file
file contained wrong changes, was commited by mistake ;
add change Ts in InvertedPendulum.cpp to allow more shooting points
casadi 3.5 uses int for mem and long longint for iw ;
remove use of nbu, nbu_idx in qp_problem, the non-constrained inputs
have to be ficticiously constrained between -inf,inf before passing to
the solver
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant