Smart Information Flow Technologies - AI planning

OpenMIND: Planning and Adapting in Domains with Novelty

musliner — Thu, 05 May 2022 18:54:28 +0000

We describe OpenMIND, a goal-oriented agent that plans to achieve its goals, executes its plans,
and revises its goals and planning models on the fly when novel, unexpected situations arise.

This general idea is not new; our contributions lie in the development of several domain-independent
model-modification and goal-reasoning heuristics that have proven effective in handling novelty,
based on blind evaluations with quantitative metrics. The metric results show that OpenMIND is
reliably able to detect novelties and then take advantage of novel opportunities and avoid or work
around novel hindrances using a small collection of domain-independent techniques. Overall, for
evaluation trials that presented novel challenges and opportunities across two different domains,
OpenMIND adapted to novelties to achieve 80% or more of the performance achieved by a baseline
agent on pre-novelty tasks in that domain.

Citation:

Proc. Ninth Conf. on Advances in Cognitive Systems

Date Published:

Friday, November 5, 2021

Publication File:

main.pdf

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages:

SIFT at ICAPS 2021

rpgoldman — Tue, 09 Nov 2021 20:39:51 +0000

SIFT researchers were active at this year's ICAPS 2021, the International Conference on Automated Planning and Scheduling. ICAPS 2021 was originally scheduled to happen in Guangzhou, in China, but had to be entirely virtual because of the COVID pandemic.

Dr. Robert Goldman was program co-chair for the conference, with Dr. Michael Katz (IBM), and Prof. Susanne Biundo (Universität Ulm). The program co-chairs coordinated the production of the proceedings, including the reviewing process, chose award recipients, and managed the program schedule.

Dr. Goldman also presented his work on integrating Monte Carlo Tree Search and HTN planning, done as part of DARPA's SD2E program, at the HTN Planning workshop, which was held in conjunction with ICAPS 2021.

Dr. Daniel Bryce gave an invited industry talk about planning for synthetic biology experimentation.

Creative Problem Solving Through Automated Planning and Analogy

musliner — Mon, 27 Jan 2020 16:30:48 +0000

As a creative problem solving strategy, analogical reasoning helps generalize and transfer solutions to new domains. Automated planners have been used for problem solving, but they reach impasses when their representation of the problem space lacks operators or resources to generate a plan from an initial condition to a set of goal conditions.

This paper presents Creative Problem Solver (CPS), a novel integration of automated planning and analogical reasoning that recognizes when analogical reasoning and generalization may resolve such impasses. CPS uses heuristics to identify missing critical resources and then uses analogical reasoning to identify suitable replacements that are readily available in the environment. We implement CPS using the CIRCA planning architecture, which supports domain-level and meta-level planning, and the structure-mapping model of analogical reasoning. This paper demonstrates CPS resolving impasses in domains where traditional automated planning would otherwise fail.

Citation:

2020 AAAI Workshop on Generalization in Planning

Date Published:

Friday, February 7, 2020

Publication File:

main.pdf

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages:

Marshal

mboldt — Tue, 17 May 2016 21:43:39 +0000

Marshal is a digital personal assistant for model-based systems. As organizations apply systems in new situations and expect new behavior, the system model may no longer characterize the environment — the model is vulnerable to drift. Marshal reins in drifting models to ensure they best match the current environment. Marshal combines user events, answers to clarifying questions, and experimentation to actively correct its knowledge of the system model. With up-to-date models, Marshal suggests corrections, maintains provenance, and predicts future model drift. Marshal reduces the cost and latency of ongoing model maintenance. SIFT has applied Marshal to maintaining NASA planning and procedure models for crew planning and Extra-Vehicular Activity (EVA).

Date Initiated:

Tuesday, July 1, 2014

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Project Full Title Hidden:

Marshal

Project Shortest Title Hidden:

Marshal

Prevent Workflow Messages:

Primary Contacts:

Dan

Maintaining Evolving Domain Models

mboldt — Tue, 17 May 2016 20:57:20 +0000

When engineering an automated planning model, domain authors typically assume a static, unchanging ground-truth world. Unfortunately, this assumption can clash with reality, where domain changes often rapidly occur in best practices, effectors, or known conditions. In these cases, remodeling the domain causes domain experts to ensure newly captured requirements integrate well with the current model. In this work, we address this model maintenance problem in a system called Marshal. Marshal assists model maintainers by reasoning about their model as a (hidden) stochastic process. It issues queries, and learns models by observing query answers, plan solutions, and direct changes to the model. Our results indicate that anticipating model evolution leads to more accurate learned models over traditional approaches.

Citation:

Daniel Bryce, J. Benton, and Michael W. Boldt. Maintaining evolving domain models. In International Joint Conference on Artificial Intelligence, Jul 2016

Date Published:

Saturday, July 9, 2016

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages:

A Hierarchical Goal-Based Formalism and Algorithm for Single-Agent Plannin

admin — Tue, 04 Aug 2015 19:16:19 +0000

Plan generation is important in a number of agent applications, but such applications generally require elaborate domain models that include not only the definitions of the actions that an agent can perform in a given domain, but also information about the most effective ways to generate plans for the agent in that domain. Such models typically take a large amount of human effort to create.

To alleviate this problem, we have developed a hierarchical goalbased planning formalism and a planning algorithm, GDP (GoalDecomposition Planner), that combines some aspects of both HTN planning and domain-independent planning. For example, it allows the planning agent to use domain-independent heuristic functions to guide the application of both methods and actions.

This paper describes the formalism, planning algorithm, correctness theorems, and the results of a large experimental study. The experiments show that our planning algorithm works as well as the well-known SHOP2 HTN planner, using domain models only about half the size of SHOP2’s.

PDF docuemnt is available at aamas-conference.org

Citation:

Shivashankar, V., Kuter, U., Nau, D., and Alford, R. A Hierarchical Goal-Based Formalism and Algorithm for Single-Agent Planning. AAMAS-12: Proceedings of the Eleventh International Conference on Autonomous Agents and Multiagent Systems, Valencia, Spain (June 2012).

Date Published:

Friday, June 1, 2012

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages: