This repository is for a currently in-development set of tools for optimization based control of robotic systems. A full description of the capabilities and documentation of the library will be available soon.
The code is written in C++ with real-time applications in mind. Ultimately, the code will be able to help users setup full order models for robots, and create, manage, and approximate constraints and costs.
Dependencies:
- Eigen 3
- Pinocchio 3.0
- OSQP & OSQP++
Officially Supported Systems:
- Ubuntu 22.04 We hope to expand the supported systems soon.
If you have installed Clarabel in a different location then these variables must be set accordingly.
If OSQP is installed properly, then the library will be automatically found and linked.
If Pinocchio is installed properly, then the library will be automatically found and linked. Note that we rely on Pinocchio 3.0, which as of 6/18/24 is only available as a build from source (but note that it is an officially supported release, not a preview branch).
When installing CppAD, make sure you use the following cmake flags: cmake -DCMAKE_BUILD_TYPE=Release -Dcppad_cxx_flags=-std=c++17
Using: https://github.com/Zolkin1/hpipm-cpp
Create the build folder. Then run
cmake --build . --config <Release/Debug/RelWithDebug> --target <target> -- -j 30
I personally like cachegrind more than perf.
Go to cmake-build-relwithdebinfo and run
valgrind --tool=memcheck ./mpc_app_test
Compile with optimizations and debug info.
Go to cmake-build-relwithdebinfo and run
valgrind --tool=cachegrind --cache-sim=yes ./mpc_app_test
To see the output:
cg_annotate cachegrind.out.<PID>
Currently appears to be encountering a segfault.
In cmake-build-relwithdebinfo run
valgrind --tool=massif ./mpc_app_test
then use ms_print.
See chat gpt
//: # (- Run with -DCMAKE_CXX_FLAGS=-fno-omit-frame-pointer)