This paper presents a unifying framework for reinforcement learning and planning.
problem Sequential decision making in AI, formalized as MDP optimization.
method A unifying algorithmic framework (FRAP) for reinforcement learning and planning.
result Identifies common dimensions in MDP planning and learning algorithms.
Reinforcement learning and symbolic planning have both been used to build intelligent autonomous agents. Reinforcement learning relies on learning from interactions with real world, which often requires an unfeasibly large amount of experience. Symbolic planning relies on manually crafted symbolic knowledge, which may …
Survey of integrating planning and learning in model-based reinforcement learning.
problem Sequential decision making in AI, formalized as MDP optimization.
method Systematic coverage of dynamics model learning and planning-learning integration.
result Broad conceptual overview of model-based reinforcement learning.
A novel pedestrian path-planning model using reinforcement learning.
problem Challenges in replicating natural pedestrian navigation behavior.
method Reinforcement learning for collision avoidance and path planning.
result Path planning by the agent closely resembles human pedestrian behavior.
Vanguard uses AI to create personalized financial plans.
problem Challenges in choosing features for complex financial planning.
method Reinforcement learning for identifying optimal savings rates.
result Trains algorithms to model financial success trajectories.
Efficient algorithms for planning in cooperative multi-agent reinforcement learning with combinatorial action spaces.
problem Planning in cooperative multi-agent reinforcement learning with a combinatorial action space.
method Efficient algorithms using local access to a simulator and linear function approximation, with improvements for additive feature decomposition and kernelized settings.
result Polynomial compute and query complexity in relevant problem parameters.
This work defines a complexity measure for BAMDP planning and introduces state abstraction for more efficient approximate planning.
problem The computational intractability of exact BAMDP planning solutions.
method Define a complexity measure for BAMDP planning, introduce state abstraction, and develop an approximate planning algorithm.
result Introduces a computationally tractable approximate planning algorithm using state abstraction.
PBCS combines RL and motion planning for better exploration.
problem RL algorithms struggle with versatile exploration in complex environments.
method PBCS uses motion planning to find a good trajectory, then trains RL on a curriculum derived from it.
result PBCS outperforms state-of-the-art RL algorithms in 2D maze environments.
The paper explores how planning with models improves credit assignment in reinforcement learning.
problem Improving credit assignment in reinforcement learning.
method Investigates the use of forward and backward models for planning in reinforcement learning.
result Establishes the relative merits and limitations of forward and backward planning mechanisms.
PALM learns abstract models for efficient planning and task transfer.
problem Efficiently learning and transferring hierarchical models for planning.
method PALM uses a new formal structure (L-AMDP) to learn independent, modular models at multiple levels of abstraction.
result PALM integrates planning and execution, facilitating rapid learning of abstract models.
New partial models correct for confounding effects in reinforcement learning.
problem Confounding effects in partial models lead to incorrect planning.
method Introduces causally correct partial models for reinforcement learning.
result Causally correct partial models avoid confounding effects and improve planning accuracy.
The paper tackles model-based RL's inaccuracy issue by dynamically adjusting planning horizons.
problem Model-based RL's failure due to model inaccuracy over long planning horizons.
method State-dependent planning horizon, learning cumulative model errors with Temporal Difference methods.
result The proposed method successfully adapts planning horizons to state-dependent model accuracy, improving policy learning efficiency.
A method uses RL to learn abstractions for planning, improving robot navigation and manipulation tasks.
problem Planning requires suitable abstractions for states and transitions, which RL struggles with for temporally extended tasks.
method Goal-conditioned policies learned with RL are incorporated into planning, with a latent variable model representing valid states.
result Our method significantly outperforms prior work on image-based robot navigation and manipulation tasks.
RL enhances LLM planning but introduces spurious solutions and diversity collapse.
problem Theoretical understanding of RL's benefits and limitations in LLM planning.
method Graph-based abstraction, policy gradient, Q-learning, supervised fine-tuning.
result RL's exploration is crucial for generalization, but PG suffers from diversity collapse.
Selective planning with imperfect models reduces harmful effects of model inadequacy.
problem Harmful effects of using an imperfect model in reinforcement learning.
method Selective planning with heteroscedastic regression to estimate predictive uncertainty from model inadequacy.
result Effective selective planning requires considering both parameter uncertainty and model inadequacy.
This paper bridges Markowitz planning and deep reinforcement learning for portfolio optimization.
problem Combining Markowitz planning and deep reinforcement learning for portfolio optimization.
method Mapping market conditions to actions using deep reinforcement learning, casting portfolio allocation as a continuous control problem.
result Deep reinforcement learning techniques can provide new insights for portfolio allocation.
Improves RL planning by proposing sub-goals hierarchically.
problem Sequential planning assumption in RL.
method Divide-and-Conquer Monte Carlo Tree Search (DC-MCTS).
result Improves navigation and control tasks.
Deep RL drone trained to compete against classical path planning in drone racing.
problem Optimizing long-term drone racing strategies using reinforcement learning.
method Used PPO algorithm on a simulated quadrotor in a racing environment created with AirSim.
result Deep RL agent outperformed classical path planning in drone racing competitions.
New method learns state embeddings from demonstrations for improved reinforcement learning.
problem Difficult relationship between observed state and useful policy actions in dynamic problems.
method Variational framework for learning state embeddings that optimize trajectory linearity.
result Learning embedding spaces improves policy gradient reinforcement learning performance.
New RL approach speeds up training across tasks.
problem Training reinforcement learning models quickly and efficiently.
method Combines planning quasi-metric and task-specific aimers.
result Achieves multiple-fold speed-up on bit-flip and robotic arm tasks.
Study shows optimal RL with transition look-ahead is NP-hard for ℓ≥2.
problem Optimal reinforcement learning with transition look-ahead is computationally hard.
method Proved NP-hardness for ℓ≥2 using linear programming. result There is a precise boundary between tractable and intractable cases for RL with look-ahead.
Proposes a method for model-based RL in complex environments without perfect simulators.
problem Lack of cheap and perfect simulators in real-world tasks.
method Induces a world program by learning dynamics and actions in graph-based environments.
result World program enables complex planning tasks in environments without perfect simulators.
Proposes learning latent reward model for planning from rewards.
problem Planning in high-dimensional state spaces with limited reward information.
method Directly learns a latent dynamics model from rewards, planning in latent state-space.
result Successfully learns accurate latent reward prediction model, achieving strong performance and high sample efficiency.
In this paper, we propose to combine imitation and reinforcement learning via the idea of reward shaping using an oracle. We study the effectiveness of the near-optimal cost-to-go oracle on the planning horizon and demonstrate that the cost-to-go oracle shortens the learner's planning horizon as function of its accurac…
The paper introduces affordances for reinforcement learning, improving planning and learning efficiency.
problem Reinforcement learning assumes all actions are available, but real-world agents face limited action spaces.
method Developed a theory of affordances for Markov Decision Processes, proposing methods to learn and use affordances.
result Affordances improve planning speed and learning efficiency, leading to simpler and better generalizing transition models.
One of the key challenges in applying reinforcement learning to real-life problems is that the amount of train-and-error required to learn a good policy increases drastically as the task becomes complex. One potential solution to this problem is to combine reinforcement learning with automated symbol planning and utili…
We introduce Imagination-Augmented Agents (I2As), a novel architecture for deep reinforcement learning combining model-free and model-based aspects. In contrast to most existing model-based reinforcement learning and planning methods, which prescribe how a model should be used to arrive at a policy, I2As learn to inter…
State-of-the-art efficient model-based Reinforcement Learning (RL) algorithms typically act by iteratively solving empirical models, i.e., by performing \emph{full-planning} on Markov Decision Processes (MDPs) built by the gathered experience. In this paper, we focus on model-based RL in the finite-state finite-horizon…
Paper tackles continual reinforcement learning by forgetting, proposing a planning method with online world models.
problem Catastrophic forgetting in reinforcement learning when learning new tasks.
method Planning with an online world model using model predictive control.
result The proposed FTL Online Agent (OA) learns new tasks without forgetting old skills.
Plan2Explore learns new tasks efficiently through self-supervised planning.
problem Challenges in reinforcement learning, especially task-specific learning and sample efficiency.
method Self-supervised exploration and fast adaptation to new tasks through efficient planning.
result Plan2Explore outperforms prior methods in learning new tasks without supervision.
The explore{exploit dilemma is one of the central challenges in Reinforcement Learning (RL). Bayesian RL solves the dilemma by providing the agent with information in the form of a prior distribution over environments; however, full Bayesian planning is intractable. Planning with the mean MDP is a common myopic approxi…
R3L uses planning algorithms to efficiently explore sparse reward environments.
problem Balancing exploration and exploitation in sparse reward reinforcement learning.
method Formulate exploration as a search problem using RRT, leverage demonstrations from initial solutions to refine RL policy.
result R3L outperforms classic and intrinsic exploration techniques, requiring fewer samples and achieving better asymptotic performance.
Plan2Vec learns image representations without labels, improving control tasks.
problem Learning image representations without labeled data.
method Constructs a weighted graph using near-neighbor distances and extrapolates to global embedding.
result Plan2Vec achieves accurate long-term value estimates in control tasks with reduced computational and memory costs.
Reward tweaking optimizes behavior for long-term goals by adjusting the reward function.
problem Optimizing behavior for long-term goals in reinforcement learning with unstable long planning horizons.
method Reward tweaking learns a surrogate reward function that induces optimal behavior for the original task.
result Reward tweaking guides agents towards better long-term returns while planning for short horizons.
Motion planning is an essential component in most of today's robotic applications. In this work, we consider the learning setting, where a set of solved motion planning problems is used to improve the efficiency of motion planning on different, yet similar problems. This setting is important in applications with rapidl…
HiDe learns hierarchical control for complex tasks by separating planning and control.
problem Solving long horizon control tasks with generalization to unseen scenarios.
method Functional decomposition of state-action spaces, RL-based planner, modular transfer of policy layers.
result Generalizes across unseen test environments and scales to longer horizons.
Graph neural network executes value iteration for flexible environments.
problem Value iteration in flexible environments with direct supervision.
method Graph Neural Network (GNN) executing value iteration algorithm.
result GNN accurately models value iteration across diverse environments.
Estimates of predictive uncertainty are important for accurate model-based planning and reinforcement learning. However, predictive uncertainties---especially ones derived from modern deep learning systems---can be inaccurate and impose a bottleneck on performance. This paper explores which uncertainties are needed for…
This paper surveys DRL for autonomous vehicle motion planning.
problem Designing intelligent motion planning for autonomous vehicles.
method Deep Reinforcement Learning (DRL) for hierarchical motion planning.
result Survey of state-of-the-art DRL solutions for autonomous vehicle motion planning.
Deep reinforcement learning has learned to play many games well, but failed on others. To better characterize the modes and reasons of failure of deep reinforcement learners, we test the widely used Asynchronous Actor-Critic (A2C) algorithm on four deceptive games, which are specially designed to provide challenges to …
We introduce the value iteration network (VIN): a fully differentiable neural network with a `planning module' embedded within. VINs can learn to plan, and are suitable for predicting outcomes that involve planning-based reasoning, such as policies for reinforcement learning. Key to our approach is a novel differentiab…
New algorithm improves reinforcement learning from partial observations.
problem Inferior performance of algorithms in real-world reinforcement learning due to partial observability.
method Representation-based approach to POMDPs, leading to a tractable algorithm.
result Empirically demonstrates superior performance with partial observations.
Deep RL finds efficient pathways for sugar to chemicals.
problem Finding efficient pathways from sugar to value-added chemicals.
method Markov decision process with deep reinforcement learning.
result Promising preliminary results in efficient biomass conversion.
The computational costs of inference and planning have confined Bayesian model-based reinforcement learning to one of two dismal fates: powerful Bayes-adaptive planning but only for simplistic models, or powerful, Bayesian non-parametric models but using simple, myopic planning strategies such as Thompson sampling. We …
Improved CEM for fast real-time planning in high-dimensional control tasks.
problem Sampling inefficiency of CEM in real-time planning.
method Novel additions to CEM including temporally-correlated actions and memory.
result 2.7-22x less samples and 1.2-10x performance increase.
We introduce Dynamic Planning Networks (DPN), a novel architecture for deep reinforcement learning, that combines model-based and model-free aspects for online planning. Our architecture learns to dynamically construct plans using a learned state-transition model by selecting and traversing between simulated states and…
Model-Based Offline Planning (MBOP) learns models from offline data to control systems directly.
problem Training RL policies from offline data without direct system interaction.
method Generates models from offline data and uses planning to control the system.
result Near-optimal policies found for simulated systems with minimal real-time interaction.
This research introduces an autonomous robot navigation method using reinforcement learning.
problem Improving robot navigation in complex environments.
method Deep Q Network (DQN) and Proximal Policy Optimization (PPO) models for path planning and decision-making.
result The models enhance robot navigation ability and adaptive learning in unknown environments.