KL-regularized reinforcement learning from expert demonstrations has proved successful in improving the sample efficiency of deep reinforcement learning algorithms, allowing them to be applied to challenging physical real-world tasks. However, we show that KL-regularized reinforcement learning with behavioral reference policies derived from expert demonstrations can suffer from pathological training dynamics that can lead to slow, unstable, and suboptimal online learning. We show empirically that the pathology occurs for commonly chosen behavioral policy classes and demonstrate its impact on sample efficiency and online policy performance. Finally, we show that the pathology can be remedied by non-parametric behavioral reference policies and that this allows KL-regularized reinforcement learning to significantly outperform state-of-the-art approaches on a variety of challenging locomotion and dexterous hand manipulation tasks.

Function approximation has enabled remarkable advances in applying reinforcement learning (RL) techniques in environments with high-dimensional inputs, such as images, in an end-to-end fashion, mapping such inputs directly to low-level control. Nevertheless, these have proved vulnerable to small adversarial input perturbations. A number of approaches for improving or certifying robustness of end-to-end RL to adversarial perturbations have emerged as a result, focusing on cumulative reward. However, what is often at stake in adversarial scenarios is the violation of fundamental properties, such as safety, rather than the overall reward that combines safety with efficiency. Moreover, properties such as safety can only be defined with respect to true state, rather than the high-dimensional raw inputs to end-to-end policies. To disentangle nominal efficiency and adversarial safety, we situate RL in deterministic partially-observable Markov decision processes (POMDPs) with the goal of maximizing cumulative reward subject to safety constraints. We then propose a partially-supervised reinforcement learning (PSRL) framework that takes advantage of an additional assumption that the true state of the POMDP is known at training time. We present the first approach for certifying safety of PSRL policies under adversarial input perturbations, and two adversarial training approaches that make direct use of PSRL. Our experiments demonstrate both the efficacy of the proposed approach for certifying safety in adversarial environments, and the value of the PSRL framework coupled with adversarial training in improving certified safety while preserving high nominal reward and high-quality predictions of true state.

A major goal in robotics is to enable intelligent mobile robots to operate smoothly in shared human-robot environments. One of the most fundamental capabilities in service of this goal is competent navigation in this ``social" context. As a result, there has been a recent surge of research on social navigation; and especially as it relates to the handling of conflicts between agents during social navigation. These developments introduce a variety of models and algorithms, however as this research area is inherently interdisciplinary, many of the relevant papers are not comparable and there is no shared standard vocabulary. This survey aims to bridge this gap by introducing such a common language, using it to survey existing work, and highlighting open problems. It starts by defining the boundaries of this survey to a limited, yet highly common type of social navigation - conflict avoidance. Within this proposed scope, this survey introduces a detailed taxonomy of the conflict avoidance components. This survey then maps existing work into this taxonomy, while discussing papers using its framing. Finally, this paper proposes some future research directions and open problems that are currently on the frontier of social navigation to aid ongoing and future research.

Although most reinforcement learning research has centered on competitive games, little work has been done on applying it to co-operative multiplayer games or text-based games. Codenames is a board game that involves both asymmetric co-operation and natural language processing, which makes it an excellent candidate for advancing RL research. To my knowledge, this work is the first to formulate Codenames as a Markov Decision Process and apply some well-known reinforcement learning algorithms such as SAC, PPO, and A2C to the environment. Although none of the above algorithms converge for the Codenames environment, neither do they converge for a simplified environment called ClickPixel, except when the board size is small.

Efficient exploration strategy is one of essential issues in cooperative multi-agent reinforcement learning (MARL) algorithms requiring complex coordination. In this study, we introduce a new exploration method with the strangeness that can be easily incorporated into any centralized training and decentralized execution (CTDE)-based MARL algorithms. The strangeness refers to the degree of unfamiliarity of the observations that an agent visits. In order to give the observation strangeness a global perspective, it is also augmented with the the degree of unfamiliarity of the visited entire state. The exploration bonus is obtained from the strangeness and the proposed exploration method is not much affected by stochastic transitions commonly observed in MARL tasks. To prevent a high exploration bonus from making the MARL training insensitive to extrinsic rewards, we also propose a separate action-value function trained by both extrinsic reward and exploration bonus, on which a behavioral policy to generate transitions is designed based. It makes the CTDE-based MARL algorithms more stable when they are used with an exploration method. Through a comparative evaluation in didactic examples and the StarCraft Multi-Agent Challenge, we show that the proposed exploration method achieves significant performance improvement in the CTDE-based MARL algorithms.

Multi-uncertainties from power sources and loads have brought significant challenges to the stable demand supply of various resources at islands. To address these challenges, a comprehensive scheduling framework is proposed by introducing a model-free deep reinforcement learning (DRL) approach based on modeling an island integrated energy system (IES). In response to the shortage of freshwater on islands, in addition to the introduction of seawater desalination systems, a transmission structure of "hydrothermal simultaneous transmission" (HST) is proposed. The essence of the IES scheduling problem is the optimal combination of each unit's output, which is a typical timing control problem and conforms to the Markov decision-making solution framework of deep reinforcement learning. Deep reinforcement learning adapts to various changes and timely adjusts strategies through the interaction of agents and the environment, avoiding complicated modeling and prediction of multi-uncertainties. The simulation results show that the proposed scheduling framework properly handles multi-uncertainties from power sources and loads, achieves a stable demand supply for various resources, and has better performance than other real-time scheduling methods, especially in terms of computational efficiency. In addition, the HST model constitutes an active exploration to improve the utilization efficiency of island freshwater.

We study model-based reinforcement learning (RL) for episodic Markov decision processes (MDP) whose transition probability is parametrized by an unknown transition core with features of state and action. Despite much recent progress in analyzing algorithms in the linear MDP setting, the understanding of more general transition models is very restrictive. In this paper, we establish a provably efficient RL algorithm for the MDP whose state transition is given by a multinomial logistic model. To balance the exploration-exploitation trade-off, we propose an upper confidence bound-based algorithm. We show that our proposed algorithm achieves $\tilde{\mathcal{O}}(d \sqrt{H^3 T})$ regret bound where $d$ is the dimension of the transition core, $H$ is the horizon, and $T$ is the total number of steps. To the best of our knowledge, this is the first model-based RL algorithm with multinomial logistic function approximation with provable guarantees. We also comprehensively evaluate our proposed algorithm numerically and show that it consistently outperforms the existing methods, hence achieving both provable efficiency and practical superior performance.

In addition to its public health crisis, COVID-19 pandemic has led to the shutdown and closure of workplaces with an estimated total cost of more than $16 trillion. Given the long hours an average person spends in buildings and indoor environments, this research article proposes data-driven control strategies to design optimal indoor airflow to minimize the exposure of occupants to viral pathogens in built environments. A general control framework is put forward for designing an optimal velocity field and proximal policy optimization, a reinforcement learning algorithm is employed to solve the control problem in a data-driven fashion. The same framework is used for optimal placement of disinfectants to neutralize the viral pathogens as an alternative to the airflow design when the latter is practically infeasible or hard to implement. We show, via simulation experiments, that the control agent learns the optimal policy in both scenarios within a reasonable time. The proposed data-driven control framework in this study will have significant societal and economic benefits by setting the foundation for an improved methodology in designing case-specific infection control guidelines that can be realized by affordable ventilation devices and disinfectants.

Many problems involve the use of models which learn probability distributions or incorporate randomness in some way. In such problems, because computing the true expected gradient may be intractable, a gradient estimator is used to update the model parameters. When the model parameters directly affect a probability distribution, the gradient estimator will involve score function terms. This paper studies baselines, a variance reduction technique for score functions. Motivated primarily by reinforcement learning, we derive for the first time an expression for the optimal state-dependent baseline, the baseline which results in a gradient estimator with minimum variance. Although we show that there exist examples where the optimal baseline may be arbitrarily better than a value function baseline, we find that the value function baseline usually performs similarly to an optimal baseline in terms of variance reduction. Moreover, the value function can also be used for bootstrapping estimators of the return, leading to additional variance reduction. Our results give new insight and justification for why value function baselines and the generalized advantage estimator (GAE) work well in practice.

In many sequential decision-making tasks, the agent is not able to model the full complexity of the world, which consists of multitudes of relevant and irrelevant information. For example, a person walking along a city street who tries to model all aspects of the world would quickly be overwhelmed by a multitude of shops, cars, and people moving in and out of view, each following their own complex and inscrutable dynamics. Is it possible to turn the agent's firehose of sensory information into a minimal latent state that is both necessary and sufficient for an agent to successfully act in the world? We formulate this question concretely, and propose the Agent Control-Endogenous State Discovery algorithm (AC-State), which has theoretical guarantees and is practically demonstrated to discover the minimal control-endogenous latent state which contains all of the information necessary for controlling the agent, while fully discarding all irrelevant information. This algorithm consists of a multi-step inverse model (predicting actions from distant observations) with an information bottleneck. AC-State enables localization, exploration, and navigation without reward or demonstrations. We demonstrate the discovery of the control-endogenous latent state in three domains: localizing a robot arm with distractions (e.g., changing lighting conditions and background), exploring a maze alongside other agents, and navigating in the Matterport house simulator.