EE/CS 381: Sensorimotor Learning for Embodied Agents

Overview:

This is an advanced course focusing on modern machine-learning algorithms for autonomous robots as embodied intelligent agents. It covers advanced topics that center around what embodied AI is and how it differs from internet AI. How embodied agents perceive their environment from raw sensory data, make decisions, and continually adapt to the physical world through both hardware and software improvements.  By the end of the course, we hope to prepare you for conducting independent research in this area, knowing how to formulate the problem, design the algorithm, critically validate the idea through experimental designs, and finally, clearly communicate the findings in both writing and presentation.   Students are expected to read, present, and debate the latest research papers on embodied AI, as well as obtain hands-on experience through a capstone project. 

• Instructor: Shuran Song
  OH: Mon 10:50 - 11:45 outside the lecture hall 

• TA: Zeyi Liu
  OH by appointment through Slack. 

• Time:  MW 9:30a-10:50a

• Location:  McMurtry Art Building Oshman 

Announcement:

• Add yourself to the course Slack channel (from Canvas). 

• Link to role assignment: See schedule below tab 2 "roles". 

• Link to presentation slides: see schedule below.  

• This course requires in-person attendance for all classes.

Pre-requisites:

• Familiarity with the basic linear algebra.

• Knowledge of basic machine learning, computer vision, and robotics.  

• If you are not sure whether you are ready for the course, please check the course instructor. 

Grading:

• Presentation & Discussion (60 %)

  • In class discussion (50%) Students will  discuss each paper with one of 7 assigned roles (e.g., author, reviewer, learner ...)   

  • Offline discussion (10%): participate in discussion on Slack 

• Course project (40 %)

  • Project proposal (15%)

  • Final project presentation (15%) 

  • Project write-up (10%)
    some potential ideas for the projects can be found here.
    
    

• Academic Honesty:
  Academic dishonesty may result in failing the course and the case will be reviewed by the Dean’s office.
  What are examples of academic dishonesty in this course?
  - Downloading slides and acting as if they are your own
  - Downloading code and acting as if they are your own
  If you are unsure, cite your sources.

Resources

• PyTorch tutorial

• TensorFlow

• PyBullet Quickstart Guide

• OpenAI Spinning Up in Deep RL

Not seeing anything above? Reauthenticate