Model-based reinforcement learning (MBRL) is a variant of the iterative learning framework, reinforcement learning, that includes a structured component of the system that is solely optimized to model the environment dynamics. Learning a model is broadly motivated from biology, optimal control, and more – it is grounded in natural human intuition of planning before acting. In this post, we discuss how model-based reinforcement learning is more susceptible to parameter tuning and how AutoML can help in finding very well performing parameter settings and schedules. Below, the top animation is the expected behavior of an agent maximizing velocity on a "Half Cheetah" robotic task, and underneath is what our paper with hyperparameter tuning finds. Model-based reinforcement learning (MBRL) is an iterative framework for solving tasks in a partially understood environment.

This work re-implements the OpenAI Gym multi-goal robotic manipulation environment, originally based on the commercial Mujoco engine, onto the open-source Pybullet engine. By comparing the performances of the Hindsight Experience Replay-aided Deep Deterministic Policy Gradient agent on both environments, we demonstrate our successful re-implementation of the original environment. Besides, we provide users with new APIs to access a joint control mode, image observations and goals with customisable camera and a built-in on-hand camera. We further design a set of multi-step, multi-goal, long-horizon and sparse reward robotic manipulation tasks, aiming to inspire new goal-conditioned reinforcement learning algorithms for such challenges. We use a simple, human-prior-based curriculum learning method to benchmark the multi-step manipulation tasks. Discussions about future research opportunities regarding this kind of tasks are also provided.

We study the statistical limits of uniform convergence for offline policy evaluation (OPE) problems (uniform OPE for short) with model-based methods under episodic MDP setting. Uniform OPE $\sup_\Pi|Q^\pi-\hat{Q}^\pi|<\epsilon$ (initiated by Yin et al. 2021) is a stronger measure than the point-wise (fixed policy) OPE and ensures offline policy learning when $\Pi$ contains all policies (we call it global policy class). In this paper, we establish an $\Omega(H^2 S/d_m\epsilon^2)$ lower bound (over model-based family) for the global uniform OPE, where $d_m$ is the minimal state-action distribution induced by the behavior policy. The order $S/d_m\epsilon^2$ reveals global uniform OPE task is intrinsically harder than offline policy learning due to the extra $S$ factor. Next, our main result establishes an episode complexity of $\tilde{O}(H^2/d_m\epsilon^2)$ for \emph{local} uniform convergence that applies to all \emph{near-empirically optimal} policies for the MDPs with \emph{stationary} transition. The result implies the optimal sample complexity for offline learning and separates local uniform OPE from the global case. Paramountly, the model-based method combining with our new analysis technique (singleton absorbing MDP) can be adapted to the new settings: offline task-agnostic and the offline reward-free with optimal complexity $\tilde{O}(H^2\log(K)/d_m\epsilon^2)$ ($K$ is the number of tasks) and $\tilde{O}(H^2S/d_m\epsilon^2)$ respectively, which provides a unified framework for simultaneously solving different offline RL problems.

Offline reinforcement learning (RL) has increasingly become the focus of the artificial intelligent research due to its wide real-world applications where the collection of data may be difficult, time-consuming, or costly. In this paper, we first propose a two-fold taxonomy for existing offline RL algorithms from the perspective of exploration and exploitation tendency. Secondly, we derive the explicit expression of the upper bound of extrapolation error and explore the correlation between the performance of different types of algorithms and the distribution of actions under states. Specifically, we relax the strict assumption on the sufficiently large amount of state-action tuples. Accordingly, we provably explain why batch constrained Q-learning (BCQ) performs better than other existing techniques. Thirdly, after identifying the weakness of BCQ on dataset of low mean episode returns, we propose a modified variant based on top return selection mechanism, which is proved to be able to gain state-of-the-art performance on various datasets. Lastly, we create a benchmark platform on the Atari domain, entitled RL easy go (RLEG), at an estimated cost of more than 0.3 million dollars. We make it open-source for fair and comprehensive competitions between offline RL algorithms with complete datasets and checkpoints being provided.

Scaling issues are mundane yet irritating for practitioners of reinforcement learning. Error scales vary across domains, tasks, and stages of learning; sometimes by many orders of magnitude. This can be detrimental to learning speed and stability, create interference between learning tasks, and necessitate substantial tuning. We revisit this topic for agents based on temporal-difference learning, sketch out some desiderata and investigate scenarios where simple fixes fall short. The mechanism we propose requires neither tuning, clipping, nor adaptation. We validate its effectiveness and robustness on the suite of Atari games. Our scaling method turns out to be particularly helpful at mitigating interference, when training a shared neural network on multiple targets that differ in reward scale or discounting.

Most of the recent deep reinforcement learning advances take an RL-centric perspective and focus on refinements of the training objective. We diverge from this view and show we can recover the performance of these developments not by changing the objective, but by regularising the value-function estimator. Constraining the Lipschitz constant of a single layer using spectral normalisation is sufficient to elevate the performance of a Categorical-DQN agent to that of a more elaborated \rainbow{} agent on the challenging Atari domain. We conduct ablation studies to disentangle the various effects normalisation has on the learning dynamics and show that is sufficient to modulate the parameter updates to recover most of the performance of spectral normalisation. These findings hint towards the need to also focus on the neural component and its learning dynamics to tackle the peculiarities of Deep Reinforcement Learning.

Imagine you were interested in solving a certain problem using Reinforcement learning. You have coded in your environment and you compile a laundry list of Reinforcement Learning (RL) algorithms to try. Self-implementing the algorithms from scratch is tricky and time-consuming because it requires a lot of trials and contains a lot of implementational tricks. The best answer is to use an RL framework. RL framework contains near-optimal implementations of RL algorithms.

A practical approach to robot reinforcement learning is to first collect a large batch of real or simulated robot interaction data, using some data collection policy, and then learn from this data to perform various tasks, using offline learning algorithms. Previous work focused on manually designing the data collection policy, and on tasks where suitable policies can easily be designed, such as random picking policies for collecting data about object grasping. For more complex tasks, however, it may be difficult to find a data collection policy that explores the environment effectively, and produces data that is diverse enough for the downstream task. In this work, we propose that data collection policies should actively explore the environment to collect diverse data. In particular, we develop a simple-yet-effective goal-conditioned reinforcement-learning method that actively focuses data collection on novel observations, thereby collecting a diverse data-set. We evaluate our method on simulated robot manipulation tasks with visual inputs and show that the improved diversity of active data collection leads to significant improvements in the downstream learning tasks.

This paper studies the problem of autonomous exploration under localization uncertainty for a mobile robot with 3D range sensing. We present a framework for self-learning a high-performance exploration policy in a single simulation environment, and transferring it to other environments, which may be physical or virtual. Recent work in transfer learning achieves encouraging performance by domain adaptation and domain randomization to expose an agent to scenarios that fill the inherent gaps in sim2sim and sim2real approaches. However, it is inefficient to train an agent in environments with randomized conditions to learn the important features of its current state. An agent can use domain knowledge provided by human experts to learn efficiently. We propose a novel approach that uses graph neural networks in conjunction with deep reinforcement learning, enabling decision-making over graphs containing relevant exploration information provided by human experts to predict a robot's optimal sensing action in belief space. The policy, which is trained only in a single simulation environment, offers a real-time, scalable, and transferable decision-making strategy, resulting in zero-shot transfer to other simulation environments and even real-world environments.

Efficient and robust policy transfer remains a key challenge for reinforcement learning to become viable for real-wold robotics. Policy transfer through warm initialization, imitation, or interacting over a large set of agents with randomized instances, have been commonly applied to solve a variety of Reinforcement Learning tasks. However, this seems far from how skill transfer happens in the biological world: Humans and animals are able to quickly adapt the learned behaviors between similar tasks and learn new skills when presented with new situations. Here we seek to answer the question: Will learning to combine adaptation and exploration lead to a more efficient transfer of policies between domains? We introduce a principled mechanism that can "Adapt-to-Learn", that is adapt the source policy to learn to solve a target task with significant transition differences and uncertainties. We show that the presented method learns to seamlessly combine learning from adaptation and exploration and leads to a robust policy transfer algorithm with significantly reduced sample complexity in transferring skills between related tasks.