Exploration in environments which differ across episodes has received increasing attention in recent years. Current methods use some combination of global novelty bonuses, computed using the agent's entire training experience, and \textit{episodic novelty bonuses}, computed using only experience from the current episode. However, the use of these two types of bonuses has been ad-hoc and poorly understood. In this work, we shed light on the behavior of these two types of bonuses through controlled experiments on easily interpretable tasks as well as challenging pixel-based settings. We find that the two types of bonuses succeed in different settings, with episodic bonuses being most effective when there is little shared structure across episodes and global bonuses being effective when more structure is shared. We develop a conceptual framework which makes this notion of shared structure precise by considering the variance of the value function across contexts, and which provides a unifying explanation of our empirical results. We furthermore find that combining the two bonuses can lead to more robust performance across different degrees of shared structure, and investigate different algorithmic choices for defining and combining global and episodic bonuses based on function approximation. This results in an algorithm which sets a new state of the art across 16 tasks from the MiniHack suite used in prior work, and also performs robustly on Habitat and Montezuma's Revenge.

Learning fine-grained movements is a challenging topic in robotics, particularly in the context of robotic hands. One specific instance of this challenge is the acquisition of fingerspelling sign language in robots. In this paper, we propose an approach for learning dexterous motor imitation from video examples without additional information. To achieve this, we first build a URDF model of a robotic hand with a single actuator for each joint. We then leverage pre-trained deep vision models to extract the 3D pose of the hand from RGB videos. Next, using state-of-the-art reinforcement learning algorithms for motion imitation (namely, proximal policy optimization and soft actor-critic), we train a policy to reproduce the movement extracted from the demonstrations. We identify the optimal set of hyperparameters for imitation based on a reference motion. Finally, we demonstrate the generalizability of our approach by testing it on six different tasks, corresponding to fingerspelled letters. Our results show that our approach is able to successfully imitate these fine-grained movements without additional information, highlighting its potential for real-world applications in robotics.

In the same way that the computer vision (CV) and natural language processing (NLP) communities have developed self-supervised methods, reinforcement learning (RL) can be cast as a self-supervised problem: learning to reach any goal, without requiring human-specified rewards or labels. However, actually building a self-supervised foundation for RL faces some important challenges. Building on prior contrastive approaches to this RL problem, we conduct careful ablation experiments and discover that a shallow and wide architecture, combined with careful weight initialization and data augmentation, can significantly boost the performance of these contrastive RL approaches on challenging simulated benchmarks. Additionally, we demonstrate that, with these design decisions, contrastive approaches can solve real-world robotic manipulation tasks, with tasks being specified by a single goal image provided after training.

Many Deep Reinforcement Learning (D-RL) algorithms rely on simple forms of exploration such as the additive action noise often used in continuous control domains. Typically, the scaling factor of this action noise is chosen as a hyper-parameter and is kept constant during training. In this paper, we focus on action noise in off-policy deep reinforcement learning for continuous control. We analyze how the learned policy is impacted by the noise type, noise scale, and impact scaling factor reduction schedule. We consider the two most prominent types of action noise, Gaussian and Ornstein-Uhlenbeck noise, and perform a vast experimental campaign by systematically varying the noise type and scale parameter, and by measuring variables of interest like the expected return of the policy and the state-space coverage during exploration. For the latter, we propose a novel state-space coverage measure $\operatorname{X}_{\mathcal{U}\text{rel}}$ that is more robust to estimation artifacts caused by points close to the state-space boundary than previously-proposed measures. Larger noise scales generally increase state-space coverage. However, we found that increasing the space coverage using a larger noise scale is often not beneficial. On the contrary, reducing the noise scale over the training process reduces the variance and generally improves the learning performance. We conclude that the best noise type and scale are environment dependent, and based on our observations derive heuristic rules for guiding the choice of the action noise as a starting point for further optimization.

We introduce hybrid execution in multi-agent reinforcement learning (MARL), a new paradigm in which agents aim to successfully complete cooperative tasks with arbitrary communication levels at execution time by taking advantage of information-sharing among the agents. Under hybrid execution, the communication level can range from a setting in which no communication is allowed between agents (fully decentralized), to a setting featuring full communication (fully centralized), but the agents do not know beforehand which communication level they will encounter at execution time. To formalize our setting, we define a new class of multi-agent partially observable Markov decision processes (POMDPs) that we name hybrid-POMDPs, which explicitly model a communication process between the agents. We contribute MARO, an approach that makes use of an auto-regressive predictive model, trained in a centralized manner, to estimate missing agents' observations at execution time. We evaluate MARO on standard scenarios and extensions of previous benchmarks tailored to emphasize the negative impact of partial observability in MARL. Experimental results show that our method consistently outperforms relevant baselines, allowing agents to act with faulty communication while successfully exploiting shared information.

In multi-agent reinforcement learning, the behaviors that agents learn in a single Markov Game (MG) are typically confined to the given agent number. Every single MG induced by varying the population may possess distinct optimal joint strategies and game-specific knowledge, which are modeled independently in modern multi-agent reinforcement learning algorithms. In this work, our focus is on creating agents that can generalize across population-varying MGs. Instead of learning a unimodal policy, each agent learns a policy set comprising effective strategies across a variety of games. To achieve this, we propose Meta Representations for Agents (MRA) that explicitly models the game-common and game-specific strategic knowledge. By representing the policy sets with multi-modal latent policies, the game-common strategic knowledge and diverse strategic modes are discovered through an iterative optimization procedure. We prove that by approximately maximizing the resulting constrained mutual information objective, the policies can reach Nash Equilibrium in every evaluation MG when the latent space is sufficiently large. When deploying MRA in practical settings with limited latent space sizes, fast adaptation can be achieved by leveraging the first-order gradient information. Extensive experiments demonstrate the effectiveness of MRA in improving training performance and generalization ability in challenging evaluation games.

Goal-conditioned reinforcement learning (RL) with sparse rewards remains a challenging problem in deep RL. Hindsight Experience Replay (HER) has been demonstrated to be an effective solution, where HER replaces desired goals in failed experiences with practically achieved states. Existing approaches mainly focus on either exploration or exploitation to improve the performance of HER. From a joint perspective, exploiting specific past experiences can also implicitly drive exploration. Therefore, we concentrate on prioritizing both original and relabeled samples for efficient goal-conditioned RL. To achieve this, we propose a novel value consistency prioritization (VCP) method, where the priority of samples is determined by the consistency of ensemble Q-values. This distinguishes the VCP method with most existing prioritization approaches which prioritizes samples based on the uncertainty of ensemble Q-values. Through extensive experiments, we demonstrate that VCP achieves significantly higher sample efficiency than existing algorithms on a range of challenging goal-conditioned manipulation tasks. We also visualize how VCP prioritizes good experiences to enhance policy learning.

Although Reinforcement Learning (RL) has shown to be capable of producing impressive results, its use is limited by the impact of its hyperparameters on performance. This often makes it difficult to achieve good results in practice. Automated RL (AutoRL) addresses this difficulty, yet little is known about the dynamics of the hyperparameter landscapes that hyperparameter optimization (HPO) methods traverse in search of optimal configurations. In view of existing AutoRL approaches dynamically adjusting hyperparameter configurations, we propose an approach to build and analyze these hyperparameter landscapes not just for one point in time but at multiple points in time throughout training. Addressing an important open question on the legitimacy of such dynamic AutoRL approaches, we provide thorough empirical evidence that the hyperparameter landscapes strongly vary over time across representative algorithms from RL literature (DQN, PPO, and SAC) in different kinds of environments (Cartpole, Bipedal Walker, and Hopper) This supports the theory that hyperparameters should be dynamically adjusted during training and shows the potential for more insights on AutoRL problems that can be gained through landscape analyses. Our code can be found at https://github.com/automl/AutoRL-Landscape

We present an information-theoretic framework to learn fixed-dimensional embeddings for tasks in reinforcement learning. We leverage the idea that two tasks are similar to each other if observing an agent's performance on one task reduces our uncertainty about its performance on the other. This intuition is captured by our information-theoretic criterion which uses a diverse population of agents to measure similarity between tasks in sequential decision-making settings. In addition to qualitative assessment, we empirically demonstrate the effectiveness of our techniques based on task embeddings by quantitative comparisons against strong baselines on two application scenarios: predicting an agent's performance on a test task by observing its performance on a small quiz of tasks, and selecting tasks with desired characteristics from a given set of options.

This paper investigates how deep multi-agent reinforcement learning can enable the scalable and privacy-preserving coordination of residential energy flexibility. The coordination of distributed resources such as electric vehicles and heating will be critical to the successful integration of large shares of renewable energy in our electricity grid and, thus, to help mitigate climate change. The pre-learning of individual reinforcement learning policies can enable distributed control with no sharing of personal data required during execution. However, previous approaches for multi-agent reinforcement learning-based distributed energy resources coordination impose an ever greater training computational burden as the size of the system increases. We therefore adopt a deep multi-agent actor-critic method which uses a \emph{centralised but factored critic} to rehearse coordination ahead of execution. Results show that coordination is achieved at scale, with minimal information and communication infrastructure requirements, no interference with daily activities, and privacy protection. Significant savings are obtained for energy users, the distribution network and greenhouse gas emissions. Moreover, training times are nearly 40 times shorter than with a previous state-of-the-art reinforcement learning approach without the factored critic for 30 homes.