Dynamics and Domain Randomized Gait Modulation with Bezier Curves for Sim-to-Real Legged Locomotion

I'm happy to announce that my co-authors (Ian Abraham, Matthew L. Elwin, Todd D. Murphey) and I have just completed our paper: Dynamics and Domain Randomized Gait Modulation with Bezier Curves for Sim-to-Real Legged Locomotion. All of our methods and findings were conducted on OpenQuadruped using this environment, which I will be continually supporting and keeping open-source. For this reason, I've transferred this repo's ownership to my co-run organization OpenQuadruped with Adham so that we may consolidate our efforts more effectively!

To learn more, and to replicate our method yourselves, visit our website!

Also, keep the videos coming! It's great to see people building and operating OpenQuadruped!

ROSSerial Integration!

ROSSerial Integration

Thanks to a tip from Juan Miguel Jimeno - creator of CHAMP, I decided to forego custom Serial messaging in favor or ROSSerial. As a result, the Teensy firmware is cleaner, more compact, and more modular. Note that I am using ROSSerial for ROS Melodic, so follow the tutorial accordingly.

The Gait Planner (which can optionally run the RL agent) sends either joint angles or HipToFoot Vectors on the spot/joints topic, which is read by the Teensy. The Teensy publishes contact sensor and IMU data to spot/contact and spot/imu, which the RL agent can use as its state observations.

I also modified the servo_calibrator node. It will send your desired joint angle in degrees to the requested joint 0-11 and set all others to -999. The Teensy will recognize this invalid command and will not update the desired motor angle if it sees it. A how-to will be coming near the end of August.

New Design!

I'm happy to announce that together with Adham Elarabawy, I have a completed a total mechanical redesign of SpotMicro. We call it OpenQuadruped! Check out the STEP files here!

Main improvements:

• Shortened the body by 40mm while making more room for our electronics with adapter plates.
• Moved all the servos to the hip to save 60g on the lower legs, which are now actuated using belt-drives.
• Added support bridge on hip joint for added longevity.
• Added flush slots for hall effect sensors on the feet.

I also went ahead and created a new URDF with proper inertial values on each link, making the simulation much more reliable.

You can still use the original simulation in the spotmicroai branch of this repo.

Spot OpenAI Gym Environment

First release! Currently, I provide the ability to test out your own gaits (or to use my Bezier implementation) using env_tester.py. You may also use spot_ars.py and spot_ars_eval.py as a template for training and testing scripts respectively. Feel free to check out the various RL agent libraries (ars_lib, td3_lib, sac_lib, ppo_lib)for use, or add your own via Pull Request!

There is also a ROS implementation for joystick integration.

Documentation: https://spot-mini-mini.readthedocs.io/en/latest/?badge=latest