Kino-dynamic Trajectory Optimization for Multiped Robots

This software provides methods for kino-dynamic optimization for multiped robots. It contains a list of solvers and optimization problem for robotics optimal control.

Install required packages

Please make sure to install the following packages:

• the qp solver quadprog, the code manager treep and the python resource manager:

pip install lightargs==1.22 quadprog treep importlib_resources

• yaml-cpp for reading the parameters of the optimal control problem:

  • ubuntu:

  sudo apt install libyaml-cpp-dev
  

  • MacOs

  brew install yaml-cpp
  

• And colcon to build packages (more info about the build system https://design.ros2.org/articles/build_tool.html)

  • Ubuntu

  sudo apt install python3-colcon-common-extensions
  

  • MacOs

  python3 -m pip install colcon-common-extensions
  

Getting started

Clone the repository in the desired work folder <work_folder>

mkdir devel
cd devel
git clone https://github.com/machines-in-motion/treep_machines_in_motion.git
treep --clone KINO_DYN_PLANNER

This operation should have cloned pacakges in workspace/src/ which will be refered as the <work_folder>
Compile the code, by running the following commands in the <work_folder>

cd workspace
colcon build --cmake-args -DCMAKE_BUILD_TYPE=Release

Once the code has been compiled, you can source the setup.bash file in install/setup.bash

. install/setup.bash

Running a demo

cd <work_folder>/kino_dynamic_opt/momentumopt/demos
python3 ../nodes/kino_dyn_planner_solo -i <path_to_config_file>

For example plan and execute a jumping motion with

python3 ../nodes/kino_dyn_planner -i ../config/cfg_solo8_jump.yaml
python3 ../nodes/kino_dyn_planner -i ../config/cfg_solo12_jump.yaml --solo12
python3 ../nodes/kino_dyn_planner -i ../config/cfg_bolt_jump.yaml --bolt

Configuration overview

Different configuration files are available in

<work_folder>/kino_dynamic_opt/momentumopt/config

The available motions are:

• cfg_solo12_jump.yaml
• cfg_bolt_jump.yaml
• cfg_solo8_jump.yaml
• cfg_solo8_fast_short_trot.yaml
• cfg_solo8_trot.yaml

For an explanation of the different settings in the configuration files, refer to [cfg_solo_jump.yaml].

Select your desired inverse kinematics formulation

We have two different inverse kinematics formulations implemented, you can choose which one you would like to use. The default version is a first order inverse kinematics that optimizes for generalized velocities. If you wanna use second order inverse kinematics, you would need to change the boolean use_second_order_inv_kin to True in your config (.yaml) file.

Experimental Section

#### Configuration overview
The available motions are:
* Lifting rear legs: cfg_demo01_twofeet.yaml

## Saving .dat file for dynamic graph

Executing

python3 ../nodes/kino_dyn_planner -i <path_to_config_file>

automatically saves the `quadruped_positions.dat` and the `quadruped_velocities.dat` files in the `<work_folder>/kino-dynamic-opt/src/catkin/motion_planning/momentumopt/demos` directory. These files can then be supplied to dynamic graph and for example executed on the real robot.

## License

Copyright (c) 2019, New York University and Max Planck Gesellschaft.

## Authors

- Brahayam Ponton (<[email protected]>) (1)
- Majid Khadiv (<[email protected]>) (1)
- Julian Viereck (<[email protected]>) (1)
- Avadesh Meduri (<[email protected]>) (1-2)

(1): Max Planck Institute for Intelligent System, Tubingen, Germany \
(2): New York University, New York, USA