Transplanted the official teb source code to achieve common optimization and multi-path optimization.

Refer to the ros source code for transplantation.

Which can run without the ros framework and use opencv for displaying.

And replace g2o with ceres-solver by google for single path optimization.

• C++17 compiler (tested on gcc 9.1 and gcc 13.2)
• yaml-cpp (0.7.0)
• Eigen3 3.4
• Boost 1.81
• g2o (2020-04 version important!)
• ceres-solver 2.1.0
• OPENCV (4.8.0)

My operating system is archlinux. Also tested on ubuntu 20.04 in docker.

mkdir build
cd build
cmake ..
make -j16
./teb

All configurations are in the file config.yaml

• If compile reports a lot errors because of g2o then you need check your g2o version and if cmake find the g2o correctly.
• If opencv throw error that cannot use createButton then you can set show_button : false in file config.yaml.

If you have other problems, check out your libs version.

• Remove ros dependence and framework.

• Add opencv displaying.

• Replace g2o with ceres-solver. (Only support single_path and add partial redisual.)

• Support multi-path optimization with ceres-solver.

Paper:

Code :

https://github.com/rst-tu-dortmund/teb_local_planner

https://github.com/linyicheng1/teb_local_planner