Research

What if we could teach robots using language?

This research question is being tackled by the field of Interactive Task Learning. My focus is on creating robotic agents that can learn new tasks in one-shot from natural language instruction. My research involves a cognitive architecture approach, where we investigate issues of knowledge representation, reasoning, and explanation-based learning.

Rosie

Rosie is the name of the Interactive Task Learning agent I have helped develop at the University of Michigan in the Soar Lab. It is written in the Soar Cognitive Architecture. It can learn many different tasks, games, and puzzles and has been deployed in many different real and simulated environments. Here is a demo video of it learning tasks in a mobile environment.

Domains

Publications

Using Domain Knowledge to Correct Anchoring Errors in a Cog. Arch.

2019 | Aaron Mininger, John E. Laird Mininger, Laird

Demonstrates how an agent in a symbolic cognitive architecture with access to a spatial short-term memory can use spatial reasoning and domain knowledge to participate in the anchoring process and to detect and correct bottom-up anchoring errors in its input.

Characteristics of the Learning Problem in Situated Interactive Task Learning

2019 | John E. Laird, Shiwali Mohan, James Kirk, Aaron Mininger Laird, Mohan, Kirk, Mininger

Identifies and describes the key characteristics of the learning problem inherent to Situated Interactive Task Learning.

Interactively Learning a Blend of Goal and Procedural Tasks

2018 | Aaron Mininger, John E. Laird Mininger, Laird

Explains how the task-learning part of Rosie was extended to also learn simple procedural tasks in an integrated manner.

Grounding Language for Interactive Task Learning

2017 | Peter Lindes, Aaron Mininger, James Kirk, John E. Laird Lindes, Mininger, Kirk, Laird

Describes a new language comprehension system called Lucia based on Embodied Construction Grammar that is integrated with Rosie.

Interactively Learning Strategies for Handling References to Unseen or Unknown Objects

2016 | Aaron Mininger, John E. Laird Mininger, Laird

Describes work done to extend Rosie to a mobile environment and how it handles references to objects that are not visible. Rosie learns different strategies for finding objects through instruction.

Learning Task Goals Interactively with Visual Demonstrations

2016 | James Kirk, Aaron Mininger, John E. Laird Kirk, Mininger, Laird

Describes work that allows Rosie to learn task goals through visual demonstrations instead of linguistic descriptions. This was done with games, puzzles, and regular tasks on the tabletop robot.

Agent Requirements for Effective and Efficient Task-Oriented Dialog

2015 | Shiwali Mohan, James Kirk, Aaron Mininger, John E. Laird Mohan, Kirk, Mininger, Laird

Identifies and describes nine system-level requirements for collaborative robotic agents that help them support more effective, efficient, and task-oriented dialog.

Towards an Indexical Model of Situated Language Comprehension

2014 | Shiwali Mohan, Aaron Mininger, John E. Laird Mohan, Mininger, Laird

Describes a computational model of situated language comprehension that incorporates knowledge from perception, domain knowledge, and short and long term experiences when generating semantic representations.

Learning Grounded Language through Situated Interactive Instruction

2012 | Shiwali Mohan, Aaron Mininger, James Kirk, John E. Laird Mohan, Mininger, Kirk, Laird

Describes an early version of the Rosie project with the tabletop blocks-world domain that can learn nouns (shapes), adjectives (colors), prepositions (spatial relations), and verbs (goal-based actions).

Acquiring Grounded Representations of Words with Situated Interactive Instruction

2012 | Shiwali Mohan, Aaron Mininger, James Kirk, John E. Laird Mohan, Mininger, Kirk, Laird

Describes an early version of the Rosie project with the tabletop blocks-world domain that can learn nouns (shapes), adjectives (colors), prepositions (spatial relations), and verbs (goal-based actions).