HOPPER: a hierarchical planning agent for unpredictable domains
نویسندگان
چکیده
Hierarchical Task Networks (HTNs) are a family of powerful planning algorithms that have been successfully applied to many complex, real-world domains. However, they are limited to predictable domains. In this paper we present HOPPER (Hierarchical Ordered Partial-Plan Executor and Re-planner), a hierarchical planning agent that produces partial plans in a similar way to HTNs but can also handle unexpected events in unpredictable domains by interleaving planning and execution. HOPPER can detect and recover from unexpected events that invalidate the plan, and it can detect and exploit unexpected opportunities both serendipitously and by interleaving decompositions.
منابع مشابه
A Multi-Agent Extension of Hierarchical Task Network
Describing planning domains using a common formalism promotes greater reuse of research, allowing a fairer comparison between different approaches. Common planning formalisms for single-agent planning are already well established (e.g., PDDL, STRIPS, and HTN), but currently there is a shortage of multi-agent planning formalisms with clear semantics. In this paper, we propose a multi-agent exten...
متن کاملStrong Cyclic Planning with Incomplete Information and Sensing
Incomplete information and sensing are needed in order to design agents that operate in domains where their information acquisition capabilities are restricted and the environment may evolve in unpredictable ways. These representation issues, that have been studied by the work on reasoning about actions, are also being addressed from a planning perspective. The aim of this paper is to present a...
متن کاملA Replanning Algorithm for Decision Theoretic Hierarchical Planning: Principles and Empirical Evaluation
In this paper, we present a replanning algorithm for a decision-theoretic hierarchical planner, and illustrate the experimental methodology we designed to investigate its performance. The methodology relies on an agentbased framework, in which plan failures emerge from the interplay of the agent and the environment. Given this framework, the replanning algorithm is compared with planning from s...
متن کاملIntegrating Top-Down Planning with a Bottom-Up Approach that Learns to Locally Combine Actions
This paper describes two open problems in planning and presents preliminary ideas that could lead to potential solutions. The problems on which the paper focuses are (1) combining top-down and bottom-up reasoning to create hierarchical plans, and (2) representing domain knowledge in domains that are unpredictable or too complex to described or reasoned about precisely. The first problem is part...
متن کاملAssistant Agents For Sequential Planning Problems
The problem of optimal planning under uncertainty in collaborative multi-agent domains is known to be deeply intractable but still demands a solution. This thesis will explore principled approximation methods that yield tractable approaches to planning for AI assistants, which allow them to understand the intentions of humans and help them achieve their goals. AI assistants are ubiquitous in vi...
متن کامل