Robotics Seminar - Bradley Hayes - University of Colorado Boulder

Explore a robotics seminar featuring Dr. Bradley Hayes from the University of Colorado Boulder, focusing on Explainable AI for achieving shared expectations in human-robot collaboration. Delve into the challenges of deploying collaborative robots in human-dominated environments and learn about novel approaches to create adaptive, communicative robot collaborators. Discover the importance of explainability and human-interpretable models in establishing shared expectations between humans and robots, ensuring safe and efficient operation in learning from demonstration and collaborative task execution. Gain insights into Dr. Hayes' research on developing explainable AI and interpretable machine learning techniques for safe task and motion planning, aimed at creating trustworthy autonomous teammates that enhance human performance across various domains including healthcare, domestic tasks, and manufacturing.

Introduction
Explanationable AI
Shared Expectations
Classification and Interpretation
Task Execution
Demonstrations
Learning from demonstration
Learning from demonstration pipeline
Treebased demonstrations
Takeaways
Algorithm
Collaborative Robotics
Query Analysis
Key Takeaway
Using Robots to Shape Human Behavior
Learning from Environment
Compounding State Vector
Tracking Belief
Communication
Pseudocoup
Rules
Experiment
Hypothesis
Results
Sentimental Intelligence
Motivation
Issues
Summary
Feedback
Policy elicitation
Conclusion
Natural Language Understanding
Humanism
Not Talking
Out of Field
Statespace
Enable
Exaggeration
Machine Learning