Communication with computers

Communicating with Computers (CwC) Link to heading

Communication With Computers(aka CwC) started from 2015. Our focus on this project is to build a symmetric semi automated system where human and machine can semlessly collaborate to solve a problem. We are working on the specific problem of Blocks World with a collaboration with UPenn, UIUC, and UTD. The paper (Narayan-Chen et al., 2017) was published at RoboNLP workshop at ACL 2017 which contains the overall framework of the project.

Besides, we as a AI Planning team also published (Das et al., 2017) at AAAI workshop on Human-Machine Collaborative Learning and recent AAMAS (missing reference) paper tried several approaches to solve the planning aspect of the project. A more detailed version has been accepted at KBS (Das et al., 2019).

References Link to heading

  1. Planning with actively eliciting preferences Das, Mayukh, Odom, Phillip, Islam, Md. Rakibul, Doppa, Janardhan Rao, Roth, Dan, and Natarajan, Sriraam In Knowledge-Based Systems 2019

    Planning with preferences has been employed extensively to quickly generate high-quality plans. However, it may be difficult for the human expert to supply this information without knowledge of the reasoning employed by the planner. We consider the problem of actively eliciting preferences from a human expert during the planning process. Specifically, we study this problem in the context of the Hierarchical Task Network (HTN) planning framework as it allows easy interaction with the human. We propose an approach where the planner identifies when and where expert guidance will be most useful and seeks expert’s preferences accordingly to make better decisions. Our experimental results on several diverse planning domains show that the preferences gathered using the proposed approach improve the quality and speed of the planner, while reducing the burden on the human expert.

  2. Active Preference Elicitation for Planning Das, Mayukh, Islam, Md Rakibul, Doppa, Janardhan Rao, Roth, Dan, and Natarajan, Sriraam In AAAI Workshop on Human-Machine Collaborative Planning 2017

  3. Towards Problem Solving Agents that Communicate and Learn Narayan-Chen, Anjali, Graber, Colin, Das, Mayukh, Islam, Md Rakibul, Dan, Soham, Natarajan, Sriraam, Doppa, Janardhan Rao, Hockenmaier, Julia, Palmer, Martha, and Roth, Dan In RoboNLP@ACL 2017