Gait-phase Estimation Module 2 (GEM2) for Humanoid Robot Walking. The code is open-source (BSD License). Please note that this work is an on-going research and thus some parts are not fully developed yet. Furthermore, the code will be subject to changes in the future which could include greater re-factoring.
GEM2 is a machine learning framework that combines effectively unsupervised and supervised learning in a semi-supervised setting to facilitate accurate prediction/classification of the gait phase during locomotion based solely on proprioceptive sensing.
GEM2 can be used for real-time gait phase estimation. The latter functionality facilitates 3D-base/CoM estimation with the State Estimation for RObot Walking (SEROW) framework (https://github.com/mrsp/serow).
GEM2 State Publisher (https://github.com/mrsp/gem2_state_publisher)
A ROS - C/C++ package for gathering all necessary data for GEM2 in real-time.
Solely proprioceptive sensing is utilized in training, namely joint encoder, F/T, and IMUs.
These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.
- Ubuntu 16.04 and later
- ROS kinetic and later
- Sklearn
- tensorflow
- tested on python (2.7.12, 2.7.17) and python3 (3.5.2, 3.6.9)
- pip install sklearn
- pip install tensorflow (or pip install tensorflow-gpu if an nvidia gpu is installed)
- git clone https://github.com/mrsp/gem2.git
- catkin_make
- If you are using catkin tools run: catkin build
- Save the corresponding files in a similar form as in GEM2_training, GEM2_validation
- train: python train.py ../config/gem2_params_your_robot.yaml
- configure appropriately the config yaml file (in config folder) with the corresponding topics
- roslaunch gem2_ros gem2_ros.launch
The estimated GEM2 gait-phase is employed in kinematic-inertial base state estimation with SEROW (https://github.com/mrsp/serow)
