Here you will find a list of our publicly available software.
D2C-2.0 [code]: D2C-2.0 is an alternative to D2C that trains the open-loop portion of the policy using a second order method called ILQR. It is currently interfaced with MuJoCo simulator. You can use the packages to train your models (specified in ‘xml’ format). Examples are demonstrated on standard MuJoCo examples such as Pendulum, Cartpole, Acrobot, Swimmer (3-link and 6-link), Fish etc.
D2C: We have two implementations of our algorithm “D2C”, one in C++ and the other one in Python3. It is currently interfaced with MuJoCo simulator. You can use the packages to train your models (specified in ‘xml’ format). Examples are demonstrated on standard MuJoCo examples such as Pendulum, Cartpole, Swimmer (3-link and 6-link), Fish etc. The C++ library also has its implementation on one of our custom made tensegrity models.
- D2C [code]: C++ library for training the open-loop and closed-loop systems.
- D2C [code]: Python counterpart of the above.
Pose-Graph SLAM in ROS: We have shared two implementations of pose-graph SLAM in ROS using open_karto as a front-end and popular back-end solvers like G2O and GTSAM. You can use these packages to do real-time map building using laser scan data and odometry generated by a simulated robot in Gazebo or a physical robot.
- slam_karto_g2o [code]: pose-graph SLAM using g2o as backend.
- slam_karto_gtsam [code]: for pose-graph SLAM using GTSAM as backend.
Belief Space Planning with OMPL [code] : We have built an implementation of Feedback Information Road Maps (FIRM) and Node-based MultiModal Motion Planning (NBM3P) with OMPL. FIRM is a multi-query approach for planning under uncertainty which is a belief-space variant of probabilistic roadmap. M3P is a planner to localize lost robots (robots with a multi-modal belief). Our goal is to make our planners open source so as to enable the motion planning community to use methods like FIRM, NBM3P for exciting new applications. Our long term vision is to upstream these features to the OMPL library.
FIRM MATLAB Toolbox [code] : This application is an implementation of Feedback Information Road Maps (FIRM) in MATLAB. There are easy to use demos with a GUI.
Feedback / Reporting Bugs
We would really appreciate feedback for our work. Please report bugs by creating an issue on the project’s github page. If you are not on github, or would not like to post publicly, then please contact Dr. Chakravorty via email (schakrav[at]tamu[dot]edu).