When AI agents don't align their actions with human values they may cause serious harm. One way to solve the value alignment problem is by including a human operator who monitors all of the agent's actions. Despite the fact, that this solution guarantees maximal safety, it is very inefficient, since it requires the human operator to dedicate all of his attention to the agent. In this paper, we propose a much more efficient solution that allows an operator to be engaged in other activities without neglecting his monitoring task. In our approach the AI agent requests permission from the operator only for critical actions, that is, potentially harmful actions. We introduce the concept of critical actions with respect to AI safety and discuss how to build a model that measures action criticality. We also discuss how the operator's feedback could be used to make the agent smarter.

Motivated by the tremendous progress we witnessed in recent years, this paper presents a survey of the scientific literature on the topic of Collaborative Simultaneous Localization and Mapping (C-SLAM), also known as multi-robot SLAM. With fleets of self-driving cars on the horizon and the rise of multi-robot systems in industrial applications, we believe that Collaborative SLAM will soon become a cornerstone of future robotic applications. In this survey, we introduce the basic concepts of C-SLAM and present a thorough literature review. We also outline the major challenges and limitations of C-SLAM in terms of robustness, communication, and resource management. We conclude by exploring the area's current trends and promising research avenues.

In designing a strategy for a multiagent interaction an agent must balance between the assumption that opponents are behaving rationally with the risks that may occur if opponents behave irrationally. Most classic game-theoretic solution concepts, such as Nash equilibrium (NE), assume that all players are behaving rationally (and that this fact is common knowledge). On the other hand, a maximin strategy plays a strategy that has the largest worst-case guaranteed expected payoff; this limits the potential downside against a worstcase and potentially irrational opponent, but can also cause us to achieve significantly lower payoff against rational opponents. In two-player zero-sum games, Nash equilibrium and maximin strategies are equivalent (by the minimax theorem), and these two goals are completely aligned. But in non-zero-sum games and games with more than two players, this is not the case. In these games we can potentially obtain arbitrarily low payoff by following a Nash equilibrium strategy, but if we follow a maximin strategy will likely be playing far too conservatively. While the assumption that opponents are exhibiting a degree of rationality, as well as the desire to limit worst-case performance in the case of irrational opponents, are both desirable, neither the Nash equilibrium nor maximin solution concept is definitively compelling on its own. We propose a new solution concept that balances between these two extremes.

There is a knowledge gap regarding which types of failures robots undergo in domestic settings and how these failures influence customer experience. We classified 10,072 customer reviews of small utilitarian domestic robots on Amazon by the robotic failures described in them, grouping failures into twelve types and three categories (Technical, Interaction, and Service). We identified sources and types of failures previously overlooked in the literature, combining them into an updated failure taxonomy. We analyzed their frequencies and relations to customer star ratings. Results indicate that for utilitarian domestic robots, Technical failures were more detrimental to customer experience than Interaction or Service failures. Issues with Task Completion and Robustness & Resilience were commonly reported and had the most significant negative impact. Future failure-prevention and response strategies should address the technical ability of the robot to meet functional goals, operate and maintain structural integrity over time. Usability and interaction design were less detrimental to customer experience, indicating that customers may be more forgiving of failures that impact these aspects for the robots and practical uses examined. Further, we developed a Natural Language Processing model capable of predicting whether a customer review contains content that describes a failure and the type of failure it describes. With this knowledge, designers and researchers of robotic systems can prioritize design and development efforts towards essential issues.

In this paper, we contribute a multi-faceted study into Pavlovian signalling -- a process by which learned, temporally extended predictions made by one agent inform decision-making by another agent. Signalling is intimately connected to time and timing. In service of generating and receiving signals, humans and other animals are known to represent time, determine time since past events, predict the time until a future stimulus, and both recognize and generate patterns that unfold in time. We investigate how different temporal processes impact coordination and signalling between learning agents by introducing a partially observable decision-making domain we call the Frost Hollow. In this domain, a prediction learning agent and a reinforcement learning agent are coupled into a two-part decision-making system that works to acquire sparse reward while avoiding time-conditional hazards. We evaluate two domain variations: machine agents interacting in a seven-state linear walk, and human-machine interaction in a virtual-reality environment. Our results showcase the speed of learning for Pavlovian signalling, the impact that different temporal representations do (and do not) have on agent-agent coordination, and how temporal aliasing impacts agent-agent and human-agent interactions differently. As a main contribution, we establish Pavlovian signalling as a natural bridge between fixed signalling paradigms and fully adaptive communication learning between two agents. We further show how to computationally build this adaptive signalling process out of a fixed signalling process, characterized by fast continual prediction learning and minimal constraints on the nature of the agent receiving signals. Our results therefore suggest an actionable, constructivist path towards communication learning between reinforcement learning agents.

In this paper, we present a novel Bayesian online prediction algorithm for the problem setting of ad hoc teamwork under partial observability (ATPO), which enables on-the-fly collaboration with unknown teammates performing an unknown task without needing a pre-coordination protocol. Unlike previous works that assume a fully observable state of the environment, ATPO accommodates partial observability, using the agent's observations to identify which task is being performed by the teammates. Our approach assumes neither that the teammate's actions are visible nor an environment reward signal. We evaluate ATPO in three domains -- two modified versions of the Pursuit domain with partial observability and the overcooked domain. Our results show that ATPO is effective and robust in identifying the teammate's task from a large library of possible tasks, efficient at solving it in near-optimal time, and scalable in adapting to increasingly larger problem sizes.

We study what dataset assumption permits solving offline two-player zero-sum Markov game. In stark contrast to the offline single-agent Markov decision process, we show that the single strategy concentration assumption is insufficient for learning the Nash equilibrium (NE) strategy in offline two-player zero-sum Markov games. On the other hand, we propose a new assumption named unilateral concentration and design a pessimism-type algorithm that is provably efficient under this assumption. In addition, we show that the unilateral concentration assumption is necessary for learning an NE strategy. Furthermore, our algorithm can achieve minimax sample complexity without any modification for two widely studied settings: dataset with uniform concentration assumption and turn-based Markov game. Our work serves as an important initial step towards understanding offline multi-agent reinforcement learning.

A new generation of increasingly autonomous and self-learning systems, which we call embodied systems, is about to be developed. When deploying these systems into a real-life context we face various engineering challenges, as it is crucial to coordinate the behavior of embodied systems in a beneficial manner, ensure their compatibility with our human-centered social values, and design verifiably safe and reliable human-machine interaction. We are arguing that raditional systems engineering is coming to a climacteric from embedded to embodied systems, and with assuring the trustworthiness of dynamic federations of situationally aware, intent-driven, explorative, ever-evolving, largely non-predictable, and increasingly autonomous embodied systems in uncertain, complex, and unpredictable real-world contexts. We are also identifying a number of urgent systems challenges for trustworthy embodied systems, including robust and human-centric AI, cognitive architectures, uncertainty quantification, trustworthy self-integration, and continual analysis and assurance.

Meanwhile, applications of RL have only begun to expand beyond these constrained game environments to more diverse and complex real-world environments such as chip design [86], chemical reaction optimization [133] and performing long-term recommendations [45]. To further progress towards these more complex real-world environments, greater alleviation of challenges currently facing RL (e.g., generalization, robustness, scalability, and safety) is needed [7, 27, 72, 108]. Moreover, we can expect that as the complexity of environments increases, the difficulty in alleviating these challenges will increase as well [27]. For the purpose of this paper, we broadly define known RL challenges as either an aptitude or alignment problem. Aptitude encompasses challenges concerned with being able to learn. Aptitude includes ideas such as robustness, the ability of RL to perform a task (e.g., asymptotic performance) and generalize within/between environments of similar complexity; scalability, the ability of RL to scale up to more complex environment; and aptness, the rate at which a RL algorithm can learn to solve a problem or achieve a desired performance level. Likewise, alignment encompasses challenges concerned with learning as intended [7, 27, 72]. The hypothetical paperclip agent [18] is a classic example of misalignment.

Existing distributed cooperative multi-agent reinforcement learning (MARL) frameworks usually assume undirected coordination graphs and communication graphs while estimating a global reward via consensus algorithms for policy evaluation. Such a framework may induce expensive communication costs and exhibit poor scalability due to requirement of global consensus. In this work, we study MARLs with directed coordination graphs, and propose a distributed RL algorithm where the local policy evaluations are based on local value functions. The local value function of each agent is obtained by local communication with its neighbors through a directed learning-induced communication graph, without using any consensus algorithm. A zeroth-order optimization (ZOO) approach based on parameter perturbation is employed to achieve gradient estimation. By comparing with existing ZOO-based RL algorithms, we show that our proposed distributed RL algorithm guarantees high scalability. A distributed resource allocation example is shown to illustrate the effectiveness of our algorithm.