Learning hierarchical task networks by observation

Nejati, Negin; Langley, Pat; Könik, Tolga

doi:10.1145/1143844.1143928

Cited by 63 publications

(52 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Different types of representations expected by planners include: action templates represented with preconditions and effects (e.g., "put-block-on-table(x y)" describes the action of moving block x from the top of block y to the table) [34], primitive skills based on actions that the agent can perform (i.e. specifying a start condition, a set of effects, and a set of executable actions) [26], or primitives with manually specified inputs and outputs (e.g., unscrew takes in a stud as input and returns a nut as output) [24]. Other work investigated symbolic descriptions for use in low-level environments for planning [17].…”

Section: A Primitives Expected By High-level Plannersmentioning

confidence: 99%

Discovering Action Primitive Granularity from Human Motion for Human-Robot Collaboration

Grigore

Scassellati

2017

Robotics: Science and Systems XIII

View full text Add to dashboard Cite

Abstract-Developing robots capable of making sense of their environment requires the ability to learn from observations. An important paradigm that allows for robots to both imitate humans and gain an understanding of the tasks people perform is that of action primitive discovery. Action primitives have been used as a representation of the main building blocks that compose motion. Automatic primitive discovery is an active area of research, with existing methods that can provide viable solutions for learning primitives from demonstrations. However, when we learn primitives directly from raw data, we need a mechanism to determine those primitives that are appropriate for the task at hand: is brushing one's teeth a suitable primitive or are the actions of grabbing the toothbrush, adding toothpaste onto it, and executing the brushing motion better suited? It is this level of granularity that is important for determining well-suited primitives for applications. Existing methods for learning primitives do not provide a solution for discovering their granularity. Rather, these techniques stop at arbitrarily chosen levels, and often use clear, repetitive actions in order to easily label the primitives. Our contribution provides a framework for discovering the appropriate granularity level of learned primitives for a task. We apply our framework to action primitives learned from a set of motion capture data obtained from human demonstrations that includes hand and object motions. This helps find a wellsuited granularity level for our task, avoiding the use of low levels that don't capture the necessary core pattern in the actions, or high levels that miss important differences between actions. Our results show that this framework is able to discover the best suited primitive granularity level for a specific application.

show abstract

Section: A Primitives Expected By High-level Plannersmentioning

confidence: 99%

Discovering Action Primitive Granularity from Human Motion for Human-Robot Collaboration

Grigore

Scassellati

2017

Robotics: Science and Systems XIII

View full text Add to dashboard Cite

show abstract

“…With a few exceptions, there is not much attention given for supporting state search in classical planning by observed plans [16], [17]. More often the observations are applied to plan recognition [18], [19] or player action prediction [20], [21].…”

Section: Related Workmentioning

confidence: 99%

A* Heuristic Based on a Hierarchical Space Model Extracted from Game Replays

Dzieńkowski

Markowska–Kaczmar

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Abstract-The paper presents a new method of building a hierarchical model of the state space. The model is extracted fully automatically from game replays that store executed plan traces. It is used by a novel approach for estimating the distance between states in a state-space graph. The estimate is applied in the A* algorithm as a heuristic function to reduce the search space. The method was validated using the game Smart Blocks. It is a testbed environment for studying methods that benefit from game replay analysis. The proposed heuristic is dedicated to difficult classical planning problems, for which problem-specific or automated heuristics are difficult to obtain.

show abstract

“…Ilghami et al [29] and Xu and Muñoz-Avila [65] proposed eager (in the form of version spaces) and lazy learning (in the form of case-based reasoning) algorithms respectively to learn the preconditions of HTN methods, given as input the hierarchical relationships between tasks, the action models, and a complete description of the intermediate states. Nejati et al [46,51] used means-end analysis to learn structures and preconditions of the input plans, assuming that a model of the tasks in the form of Horn clauses was given. Hogg et al [26] presented an algorithm, called HTN-MAKER, to learn structures by assuming that annotated tasks are given in the form of preconditions and effects (we made the same assumption in our work).…”

Section: Htn Learningmentioning

confidence: 99%

“…Xu and Muñoz-Avila [65] presented an algorithm to learn preconditions of HTN methods under the assumption that an ontology indicating relations between the objects was given. Nejati et al [46,51] developed approaches to learn the hierarchical structures that related the tasks and subtasks under the assumption that we have the ability to completely observe any world state and to formulate the skills to be learned in the form of Horn clauses. In [26], the HTN-MAKER algorithm learns the decomposition methods for Hierarchical Task Networks.…”

Section: Introductionmentioning

confidence: 99%

Learning hierarchical task network domains from partially observed plan traces

Zhuo

Muñoz-Ávila

Yang

2014

Artificial Intelligence

View full text Add to dashboard Cite

Hierarchical Task Network (HTN) planning is an effective yet knowledge intensive problem-solving technique. It requires humans to encode knowledge in the form of methods and action models. Methods describe how to decompose tasks into subtasks and the preconditions under which those methods are applicable whereas action models describe how actions change the world. Encoding such knowledge is a difficult and time-consuming process, even for domain experts. In this paper, we propose a new learning algorithm, called HTNLearn, to help acquire HTN methods and action models. HTNLearn receives as input a collection of plan traces with partially annotated intermediate state information, and a set of annotated tasks that specify the conditions before and after the tasks' completion. In addition, plan traces are annotated with potentially empty partial decomposition trees that record the processes of decomposing tasks to subtasks. HTNLearn outputs are a collection of methods and action models. HTNLearn first encodes constraints about the methods and action models as a constraint satisfaction problem, and then solves the problem using a weighted MAX-SAT solver. HTNLearn can learn methods and action models simultaneously from partially observed plan traces (i.e., plan traces where the intermediate states are partially observable). We test HTNLearn in several HTN domains. The experimental results show that our algorithm HTNLearn is both effective and efficient.

show abstract

Learning hierarchical task networks by observation

Cited by 63 publications

References 12 publications

Discovering Action Primitive Granularity from Human Motion for Human-Robot Collaboration

Discovering Action Primitive Granularity from Human Motion for Human-Robot Collaboration

A* Heuristic Based on a Hierarchical Space Model Extracted from Game Replays

Learning hierarchical task network domains from partially observed plan traces

Contact Info

Product

Resources

About