Integrated Execution (IntEx) / Goal Reasoning (GR)

Automated planners are increasingly being integrated into online execution systems.  The integration may, for example, embed a domain-independent temporal planner in a manufacturing system (e.g., the Xerox printer application) or autonomous vehicles.  The integration may resemble something more like a “planning stack” where an automated planner produces an activity or task plan that is further refined before being executed by a reactive controller (e.g., robotics).  Or, the integration may be a domain-specific policy that maps states to actions (e.g., reinforcement learning). Online learning may or may not be involved, and may include adjusting or augmenting the model, determining when to repair versus replan, learning to switch policies, etc.  A specific focus of these integrations involves online deliberation, bringing to the foreground concerns over how much computational effort planning should invest over time.  

Goal reasoning systems often assume an online planning paradigm.  An important aspect of integrated planning and acting systems is the ability to deal with a variety of situations that may arise online. A fundamental premise of goal reasoning is reconsidering goals that may no longer be relevant as situations evolve. While some work has been done on goal reasoning capabilities in integrated planning and acting systems, there is still no consensus about how a system should monitor and progress goals during execution.  

This combined 4th Integrating Planning, Acting, and Execution (IntEx) and 8th Goal Reasoning (GR) workshop aim to bring together researchers from these subfields to encourage cross-disciplinary discussion on the challenges of integrating planning with execution, emphasize the role of goal reasoning, raising awareness, promoting discussion, and encouraging cross-fertilization of ideas.

Topics

Following from the IntEx series, we welcome papers on past topics of interest including benchmarks or challenge problems for integrated execution; improving planning performance from execution experience; plan dispatching or plan executives; anytime or incremental planning; execution monitoring, comparing online planning approaches or plan merging; managing open worlds with closed-world planners; model learning from experience or determining an observation policy; policy switching or applying incremental policy adjustment.  

As part of the special focus on Goal reasoning, we specifically welcome papers from: 

Foundations of Goal Reasoning:

• Theoretical models of goal reasoning or comparisons to other models of autonomy
• Studies of implicit goals or goal reward/value functions
• Goal management: including formulation, selection, or optimization
• Integrating planning or metareasoning with goal management
• Online goal resolution (e.g., plan repair, replanning, goal deferment, re-goaling)
• Learning, evaluation, or analysis of goal reasoning systems

Goal Reasoning Systems

• Goals in self-motivated systems, hybrid systems, Belief-Desire-Intention systems, or Goal-Driven Autonomy
• Multi-agent or distributed goal management
• Demonstrations or applications of goal reasoning systems

Human Interaction & Goal Reasoning

• Interactive goal reasoning, human-machine goal reasoning, or social goal reasoning
• Conversational or narrative reasoning about goals
• Explanation and diagnosis of notable objects or events impacting goals

Registration

The IntEx/GR workshop like the main conference is online and will take place on October 21st, 8 am-2 pm eastern time.

VIRTUAL CONFERENCE CENTER:

https://gather.town/Z1SdmgDu1rDoaMlD/ICAPS2020

To enter, you need a password. Which you get by registering here.  Once you enter the IntEx/GR workshop room it will show a message that, when pressing ‘x’, you can join the zoom call.

SLACK: 

Here is the link to intEx/GR-2020 workspace. 

Schedule

8- 8:10	Opening Remarks
8:10- 9:00	Invited Talk: Temporal Goal Networks: Work in Progress. Dana Nau. [#invited-talk-nau]	Slides
9:00-9:15	Get me to Safety! Escaping from Risks using Automated Planning. Alberto Pozanco, Yolanda E-Martín, Susana Fernandez and Daniel Borrajo. [#talk-pozanco]
9:15- 9:30	Improving Online Planning and Execution by Selecting Goals with Deep Q-Learning. Carlos Núñez-Molina and Juan Fernandez-Olivares. [#talk-núñez-molina]
9:30-9:45	Metareasoning and Path Planning for Autonomous Indoor Navigation. Susan L. Epstein and Raj Korpan. [#talk-epstein]
9:45-10	Integrated Planning, Execution and Goal Reasoning for Python. Rich Levinson. [#talk-levinson]
10-10:20	Coffee break
10:20-10:35	Using Flexible Execution, Replanning, and Model Parameter Updates to Address Environmental Uncertainty for a Planetary Lander. Daniel Wang, Joseph A. Russino, Connor Basich and Steve A. Chien.  [#talk-wang]
10:35-10:50	Using a Model of Scheduler Runtime to Improve the Effectiveness of Scheduling Embedded in Execution. Sarah Bhaskaran, Jagriti Agrawal, Steve Chien and Wayne Chi. [#talk-bhaskaran]
10:50-11:05	MEXEC: An Onboard Integrated Planning and Execution Approach for Spacecraft Commanding. Martina Troesch, Faiz Mirza, Kyle Hughes, Ansel Rothstein-Dowden, Robert Bocchino, Amanda Donner, Martin Feather, Benjamin Smith, Lorraine Fesq, Brian Barker and Brian Campuzano. [#talk-troesch]
11:05-11:20	Efficient Anytime Computation and Execution of Decoupled Robustness Envelopes for Temporal Plans. Michael Cashmore, Alessandro Cimatti, Daniele Magazzeni, Andrea Micheli and Parisa Zehtabi. [#talk-cashmore]
11:20-11:35	Robust Execution of Deterministic Plans in Non-deterministic Environments. Oscar Lima, Michael Cashmore, Daniele Magazzeni, Andrea Micheli and Rodrigo Ventura. [#talk-lima]
11:35-11:50	Reasoning About Plan Robustness Versus Plan Cost for Partially Informed Agents. Sarah Keren, Sara Bernardini, Kofi Kwapong and David Parkes. [#talk-keren]
11:50-12:20	Lunch break
12:20-13:10	Invited Talk: The integration of impasse-driven planning (and learning) with execution in Soar. John Laird. [#invited-talk-laird]
13:10-13:20	Model-Based Novelty Adaptation for Open-World AI. Matthew Klenk, Wiktor Piotrowski, Roni Stern, Shiwali Mohan and Johan de Kleer. [#talk-klenk]
13:20-13:30	Anticipatory Thinking: A New Frontier for Automated Planning. Adam Amos-Blinks, Rogelio E. Cardona-Rivera, Gene Brewer and Dustin Dannenhauer. [#talk-amos-blinks]
13:30-13:40	Rebel Agents That Adapt to Goal Expectation Failures. Zahiduddin Mohammad, Michael Cox and Matthew Molineaux. [#talk-mohammad]
13:40-13:50	Multi-Agent Goal Recognition as Implicit Ad-Hoc Teamwork. Ben Wright. [#talk-wright]
13:50-14:00	Closing Remarks.

Invited Talks

Dana Nau (University of Maryland, College Park)

John Laird (University of Michigan, Ann Arbor)

Organizers

• Zohreh Dannenhauer, Knexus Research Corp., USA,
• Mak Roberts, Naval Research Laboratory, USA,  
• Tiago Vaquero, Jet Propulsion Laboratory, USA,

Program Committee