Corpus ID: 220793532

CAMPs: Learning Context-Specific Abstractions for Efficient Planning in Factored MDPs

  title={CAMPs: Learning Context-Specific Abstractions for Efficient Planning in Factored MDPs},
  author={Rohan Chitnis and Tom Silver and Beomjoon Kim and Leslie Pack Kaelbling and Tomas Lozano-Perez},
Meta-planning, or learning to guide planning from experience, is a promising approach to improving the computational cost of planning. A general meta-planning strategy is to learn to impose constraints on the states considered and actions taken by the agent. We observe that (1) imposing a constraint can induce context-specific independences that render some aspects of the domain irrelevant, and (2) an agent can take advantage of this fact by imposing constraints on its own behavior. These… Expand

Figures and Tables from this paper

Context-Specific Representation Abstraction for Deep Option Learning
This paper introduces Context-Specific Representation Abstraction for Deep Option Learning (CRADOL), a new framework that considers both temporal abstraction and context-specific representation abstraction to effectively reduce the size of the search over policy space. Expand
Discovering State and Action Abstractions for Generalized Task and Motion Planning
An algorithm for learning features, abstractions, and generalized plans for continuous robotic task and motion planning (TAMP) is proposed and the unique difficulties that arise when forced to consider geometric and physical constraints as a part of the generalized plan are examined. Expand


Combined task and motion planning through an extensible planner-independent interface layer
This work proposes a new approach that uses off-the-shelf task planners and motion planners and makes no assumptions about their implementation and uses a novel representational abstraction that requires only that failures in computing a motion plan for a high-level action be identifiable and expressible in the form of logical predicates at the task level. Expand
Guided search for task and motion plans using learned heuristics
This work presents an algorithm that searches the space of possible task and motion plans and uses statistical machine learning to guide the search process and shows significant improvements in performance over prior work. Expand
Relational envelope-based planning
This work shows how structured representations of the environment's dynamics can constrain and speed up the planning process and uses the envelope MDP framework to create a Markov decision process out of a subset of the possible state space. Expand
Learning to guide task and motion planning using score-space representation
A method that predicts constraints on the search space based on a generic representation of a planning problem instance, called score space, where it represents a problem instance in terms of performance of a set of solutions attempted so far, and transfers knowledge, in the form of constraints, from previous problems based on the similarity in score space. Expand
Proximity-Based Non-uniform Abstractions for Approximate Planning
Qualitative and quantitative results are presented showing that an agent exploiting this novel approximation method successfully finds solutions to the planning problem using much less than the full state space. Expand
Learning Feasibility for Task and Motion Planning in Tabletop Environments
The key idea is to learn a classifier for feasible motions and to use this classifier as a heuristic to order the search for a task–motion plan, which reduces the total number of motion planning attempts. Expand
Deep Visual Heuristics: Learning Feasibility of Mixed-Integer Programs for Manipulation Planning
A deep neural network is proposed that predicts the feasibility of a mixed-integer program from visual input for robot manipulation planning by using the learned network as a heuristic to guide the search over the discrete variables of the mixed- integer program. Expand
Learning value functions with relational state representations for guiding task-and-motion planning
We propose a novel relational state representation and an action-value function learning algorithm that learns from planning experience for geometric task-and-motion planning (GTAMP) problems, inExpand
Guiding Search in Continuous State-Action Spaces by Learning an Action Sampler From Off-Target Search Experience
This paper uses a Generative Adversarial Network (GAN) to represent an action sampler, and introduces a new technique, based on an importance-ratio estimation method, for using samples from a non-target distribution to make GAN learning more data-efficient. Expand
Solving large stochastic planning problems using multiple dynamic abstractions
Experimental results show that suitable module choices gain computational tractability at little cost to behavioral optimality and allow the module hierarchy to solve larger and more interesting domains than previously possible. Expand