Instructor Spring 2018: Brad Hayes


As capabilities of actuators, sensors, artificial intelligence, and machine learning techniques increase, so too does the scope of environments within which autonomous robots may be deployed and the complexity of problems they will be tasked to solve. Importantly, these increasingly capable systems will often need to be deployed into human-populated environments, necessitating algorithms that are explicitly mindful of end users, interaction partners, and disinterested bystanders alike.

This course will examine existing algorithms in robotics with a focus on human-robot interaction and collaboration, developing new human-aware adaptations to autonomous planning and decision-making systems, enabling them to efficiently operate in conjunction with human partners or supervisors. We will investigate questions such as: How can human demonstration or feedback be leveraged to accelerate skill acquisition? How can we efficiently model human behaviors, then use those models to improve teamwork? How can a robot be more transparent in its decision-making or actions to improve co-worker safety?


The course will combine lectures with student-led paper presentations, focusing both on fundamental knowledge acquisition and discussion. Short quizzes will be given before lecture begins to ensure understanding of the material for the day – these are not meant to be onerous and will be graded as {plus, check, minus}, as they are only meant to ensure the papers for the day were read. The course will culminate in a research project focused on characterizing or developing autonomous systems that operate safely and productively in the presence of (or collaboratively with) humans.

Learning Objectives

• Articulate challenges in building autonomous systems that interact with humans
• Apply machine learning techniques to enable human-robot collaboration
• Develop an understanding of computational models of verbal and non-verbal communication
• Apply learning from demonstration techniques to enable robots to acquire new capabilities and more rapidly generalize existing ones
• Effectively communicate scientific content
• Critique scientific literature with respect to experimental design and analysis
• Implement control algorithms on real robot hardware to design real autonomous systems


This course will not have a textbook, as all readings will be posted online. Readings will be drawn heavily from conference proceedings at HRI, RSS, CORL, AAAI, IJCAI, AAMAS, and other top robotics publication venues.


This is a graduate research course designed to jump-start research in human-robot interaction. The course project may involve significant programming and engineering effort, and as such a strong programming and computer science background is required. Students will be working with physical robotic systems, and are expected to be strongly motivated to develop real autonomous systems that interact in human environments.