• Corpus ID: 238634678

Planning from Pixels in Environments with Combinatorially Hard Search Spaces

@inproceedings{Bagatella2021PlanningFP,
  title={Planning from Pixels in Environments with Combinatorially Hard Search Spaces},
  author={Marco Bagatella and Miroslav Ols{\'a}k and Michal Rol'inek and Georg Martius},
  booktitle={Neural Information Processing Systems},
  year={2021}
}
The ability to form complex plans based on raw visual input is a litmus test for current capabilities of artificial intelligence, as it requires a seamless combination of visual processing and abstract algorithmic execution, two traditionally separate areas of computer science. A recent surge of interest in this field brought advances that yield good performance in tasks ranging from arcade games to continuous control; these methods however do not come without significant issues, such as… 

Figures and Tables from this paper

Challenges to Solving Combinatorially Hard Long-Horizon Deep RL Tasks

Inspired by combinatorially hard optimization problems, a set of robotics tasks which admit many distinct solutions at the high-level, but require reasoning about states and rewards thousands of steps into the future for the best performance are proposed.

Exploring Long-Horizon Reasoning with Deep RL in Combinatorially Hard Tasks

Inspired by combinatorially hard optimization problems, a set of robotics tasks which admit many distinct solutions at the high-level, but require reasoning about states and rewards thousands of steps into the future for the best performance are proposed.

References

SHOWING 1-10 OF 46 REFERENCES

Learning Neural-Symbolic Descriptive Planning Models via Cube-Space Priors: The Voyage Home (to STRIPS)

This work demonstrates how the sophisticated innate prior placed on the learning process significantly reduces the complexity of the learned representation, and reveals a connection to the graph-theoretic notion of ``cube-like graphs'', thus opening the door to a deeper understanding of the ideal properties for learned symbolic representations.

World Model as a Graph: Learning Latent Landmarks for Planning

This work proposes to learn graph-structured world models composed of sparse, multi-step transitions and devise a novel algorithm to learn latent landmarks that are scattered across the goal space as the nodes on the graph, and believes this work is an important step towards scalable planning in reinforcement learning.

Classical Planning in Deep Latent Space: Bridging the Subsymbolic-Symbolic Boundary

This paper proposes LatPlan, an unsupervised architecture combining deep learning and classical planning, and develops Action Autoencoder / Discriminator, a neural architecture which jointly finds the action symbols and the implicit action models (preconditions/effects), and provides a successor function for the implicit graph search.

Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model

The MuZero algorithm is presented, which, by combining a tree-based search with a learned model, achieves superhuman performance in a range of challenging and visually complex domains, without any knowledge of their underlying dynamics.

Path Planning using Neural A* Search

This work reformulates a canonical A* search algorithm to be differentiable and couple it with a convolutional encoder to form an end-to-end trainable neural network planner that outperformed state-of-the-art data-driven planners in terms of the search optimality and efficiency trade-off.

Deep Sequential Models for Sampling-Based Planning

We demonstrate how a sequence model and a sampling-based planner can influence each other to produce efficient plans and how such a model can automatically learn to take advantage of observations of

Robot Motion Planning in Learned Latent Spaces

L-SBMP is presented, a methodology toward computing motion plans for complex robotic systems by learning a plannable latent representation through an autoencoding network, a dynamics network, and a collision checking network, which mirror the three main algorithmic primitives of SBMP.

Visual Reinforcement Learning with Imagined Goals

An algorithm is proposed that acquires general-purpose skills by combining unsupervised representation learning and reinforcement learning of goal-conditioned policies, efficient enough to learn policies that operate on raw image observations and goals for a real-world robotic system, and substantially outperforms prior techniques.

Learning Latent Dynamics for Planning from Pixels

The Deep Planning Network (PlaNet) is proposed, a purely model-based agent that learns the environment dynamics from images and chooses actions through fast online planning in latent space using a latent dynamics model with both deterministic and stochastic transition components.

LEGO: Leveraging Experience in Roadmap Generation for Sampling-Based Planning

An algorithm LEGO is presented that addresses the problem of leveraging prior experience to generate roadmaps in sampling-based motion planning by training the CVAE with target samples that satisfy two important criteria.