This internship position is being offered through the Computing Internships course COMP4820/8830. You must follow the instructions on the Computing Internships webpage to apply.
Organisation:
We are an ANU spin out led by Prof. Jochen Renz that focuses on making AI rapidly adaptable to changes affecting its operating environment.
Project:
Task adaptation for a humanoid/quadruped robot.
This project uses NVIDIA Isaac SIM to simulate and test the performance of a specific robot capability/task on a specific humanoid robot (currently available humanoids are from 1X, Agility, Fourier Intelligence and Sanctuary) or a specific quadruped robot (currently available quadrupeds are from ANYbotics, Boston Dynamics, and Unitree) in a physical simulation environment. Each intern will pick a different humanoid or quadruped robot and a different capability that the robot needs to perform in the simulation environment.
Typically, these capabilities are pretrained and available to use in the Isaac SIM.
As part of the internship, the interns first task is to get familiar with the NVIDIA Isaac SIM environment and the specific capability and robot they have selected (~2 weeks).
The second task is to modify the operational environment of the robot in different ways, as well as to familiarize themselves with relevant machine learning methods. (~3 weeks).
The third task is to retrain the selected capability in the different environment modifications and to analyze the retraining and the resulting performance of the robot (~4 weeks).
The fourth task is to swap robot with another intern and to retrain the specific task for the new robot, and to analyze the retraining and the resulting performance of the robot (~3 weeks).
After the internship, interns are encouraged to write a paper together or individually about their results, with the help of our team.
Required/Preferred Technical Skills:
Excellent coding skills (structured programming skills, excellent algorithmic skills) required, experience with machine learning preferred
Professional/Other Skills:
Since this internship uses environments, software languages and platforms the interns are likely not familiar with, we need interns who are excellent programmers and who can quickly familiarize themselves with and master new technology. Interns should be already willing to start familiarizing themselves with the new technology before the start of the internship.
Type of Internship:
Hybrid (in-person/remote).
The internship will be unpaid.
How to Apply:
Please follow the instructions on the Computing Internships webpage.