This is a fun project I’ve always wanted to do, a legged robot. I’ve been interested in robotics for a long time, and I’ve been looking for a project to build up my ROS skills. I saw a project on Hackaday (GitHub here) a bit ago showcasing a project by some senior undergraduate engineers’s final project where they built a robot dog with software based on the Stanford Pupper. This project upgrades the mechanical joints a bit, and also includes the ROS Gazebo environment and URDF file for simulation, a ROS control node.
This means I could use Gazebo or Mujoco for reinforcement learning to train the dog. I’m hoping to design a power distribution and GPIO board to connect to a Jetson Nano. Space is scarce in the dog after the battery is added, so it will be a bit of a challenge. So far I’ve printed the parts and finished most of the mechanical assembly. I’m working on leg calibration and software integration now. The original project uses a Playstation remote to control the robot, and I’m looking into creating a ROS node that utilizes a Google Stadia controller since I have one lying around and it’s basically a bluetooth controller now.