Autonomous Machine for Inspecting Gas & Operations (AMIGO)

Summary

As a part of the AMIGO project, I am currently heading the searching control algorithm's design. Previously, I have worked on the RL deployment, ROS functionalities, and support of rapid prototyping efforts. While this research project is loosely structured, allowing for exploration into my team's interests, our end goal is to develop an autonomous robotic platform for powerplant diagnostics. Our canvas is the Unitree Go2 EDU quadruped, where we have added a Jetson AGX Orin, 2D LiDAR, ZED X Camera, Raspberry Pi, and other equipment. With our additions, the robot can currently autonomously navigate in two dimensions, leveraging Nav2. I have also successfully trained custom RL policies through Isaac Lab, and deployed them with my ROS2 deployment framework.

Current Status

One of the tasks I was given was to develop a locomotion controller to tackle the open-backed stairs found in many powerplants. While the controller is not fully operational, I have acquainted myself with RL training principles, Isaac Lab, Isaac Gym, and RL deployment in the process. As mentioned above, I developed a ROS2 deployment of the RL policies that is now being adapted to facilitate 3D navigational deployment with object avoidance for safety. While the ROS2 deployment worked for initial locomotion testing, I am now working on a deployment framework through Unitree's SDK, in order to minimize delays. The initial deployment of a high frequency hopping policy (to combat delays introduced by ROS bloat) is shown below.

This semester, our goal is tackle autonomous room searching. This will be an invaluable asset during real-world deployment, allowing us to find anomalies in dynamically changing environments. My current approach assigns a time-dependant probability distribution to the region of interest, while the robot determines the next target waypoint in a roughly random manner. The robot tends to the hiegher probability areas, while not forgetting about lower probability areas. More details and a testing video are coming soon!

Attributions

Funded by Steam Solutions, and housed by the Engineering Physics Propulsion Lab @ ERAU. Our custom implementation, mostly authored by my colleague, Jose Castelblanco, can be found here.