Active Learning of Abstract Plan Feasibility

@article{Noseworthy2021ActiveLO,
  title={Active Learning of Abstract Plan Feasibility},
  author={Michael Noseworthy and Caris Moses and Isa Brand and Sebastian Castro and Leslie Pack Kaelbling and Tomas Lozano-Perez and Nicholas Roy},
  journal={ArXiv},
  year={2021},
  volume={abs/2107.00683}
}
Long horizon sequential manipulation tasks are effectively addressed hierarchically: at a high level of abstraction the planner searches over abstract action sequences, and when a plan is found, lower level motion plans are generated. Such a strategy hinges on the ability to reliably predict that a feasible low level plan will be found which satisfies the abstract plan. However, computing Abstract Plan Feasibility (APF) is difficult because the outcome of a plan depends on real-world phenomena… 
Inventing Relational State and Action Abstractions for Effective and Efficient Bilevel Planning
Effective and efficient planning in continuous state and action spaces is fundamentally hard, even when the transition model is deterministic and known. One way to alleviate this challenge is to
Failure is an option: Task and Motion Planning with Failing Executions
TLDR
This work proposes a framework to address a task and motion planning setting where actions can fail during execution and results in physics-based simulation indicate that the method outperforms baseline replanning strategies to deal with failing executions.
Graph-based Reinforcement Learning meets Mixed Integer Programs: An application to 3D robot assembly discovery
TLDR
This work tackles the problem of building arbitrary, predefined target structures entirely from scratch using a set of Tetris-like building blocks and a robotic manipulator and aims at efficiently decomposing the overall task into three feasible levels that benefit mutually from each other.
Beyond Pick-and-Place: Tackling Robotic Stacking of Diverse Shapes
TLDR
This work proposes a challenging and diverse set of such objects that was fully designed to require strategies beyond a simple “pick-and-place” solution and uses a vision-based reinforcement learning (RL) approach combined with interactive policy distillation and simulation-to-reality transfer.
Object-Factored Models with Partially Observable State
TLDR
This work introduces a latent-variable model of object-factored dynamics that allows a robot to rapidly generalize to new objects by using information theoretic active learning and highlights the benefits of the deep ensemble for robust performance in downstream tasks.

References

SHOWING 1-10 OF 40 REFERENCES
Active Model Learning and Diverse Action Sampling for Task and Motion Planning
TLDR
This work uses Gaussian process methods for learning the conditions of operator effectiveness from small numbers of expensive training examples collected by experimentation on a robot, and develops adaptive sampling methods for generating diverse elements of continuous sets during planning for solving a new task, so that planning is as efficient as possible.
Learning Feasibility for Task and Motion Planning in Tabletop Environments
TLDR
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.
Deep Visual Heuristics: Learning Feasibility of Mixed-Integer Programs for Manipulation Planning
TLDR
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.
Scalable lazy SMT-based motion planning
TLDR
This work shows how to generate succinct explanations for the infeasibility of a discrete plan by exploiting a relaxation of the convex program that allows detecting the earliest possible occurrence of an infeasible transition between workspace regions.
Robot planning in the space of feasible actions: two examples
  • Sanjiv Singh, A. Kelly
  • Business
    Proceedings of IEEE International Conference on Robotics and Automation
  • 1996
TLDR
This paper uses a methodology that selects between the feasible actions that a robot might execute, in effect, circumventing many of the problems faced by configuration space planners and presents two examples of working systems that employ this methodology.
A Probabilistic Framework for Constrained Manipulations and Task and Motion Planning under Uncertainty
TLDR
The proposed framework enables a robot to prioritize various interaction modes and to acquire interesting behaviors such as contact exploitation for uncertainty reduction, eventually providing a composite control scheme that is reactive to disturbance.
Learning composable models of parameterized skills
TLDR
This paper focuses on learning models of the preconditions and effects of new parameterized skills, in a form that allows those actions to be combined with existing abilities by a generative planning and execution system.
Self-Supervised Exploration via Disagreement
TLDR
This paper proposes a formulation for exploration inspired by the work in active learning literature and trains an ensemble of dynamics models and incentivizes the agent to explore such that the disagreement of those ensembles is maximized, which results in a sample-efficient exploration.
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, in
Multi-bound tree search for logic-geometric programming in cooperative manipulation domains
  • Marc Toussaint, M. Lopes
  • Computer Science
    2017 IEEE International Conference on Robotics and Automation (ICRA)
  • 2017
TLDR
This work proposes a new approximate solver scheme, combining ideas from branch-and-bound and MCTS and exploiting multiple levels of bounds to better direct the search.
...
1
2
3
4
...