The ability to form complex plans based on raw visual input is a litmus test for current capabilities of artificial intelligence, as it requires a seamless combination of visual processing and abstract algorithmic execution, two traditionally separate areas of computer science. A recent surge of interest in this field brought advances that yield good performance in tasks ranging from arcade games to continuous control; these methods however do not come without significant issues, such as limited generalization capabilities and difficulties when dealing with combinatorially hard planning instances. Our contribution is two-fold: (i) we present a method that learns to represent its environment as a latent graph and leverages state reidentification to reduce the complexity of finding a good policy from exponential to linear (ii) we introduce a set of lightweight environments with an underlying discrete combinatorial structure in which planning is challenging even for humans. Moreover, we show that our methods achieves strong empirical generalization to variations in the environment, even across highly disadvantaged regimes, such as "one-shot" planning, or in an offline RL paradigm which only provides low-quality trajectories.

Reasoning at multiple levels of temporal abstraction is one of the key attributes of intelligence. In reinforcement learning, this is often modeled through temporally extended courses of actions called options. Options allow agents to make predictions and to operate at different levels of abstraction within an environment. Nevertheless, approaches based on the options framework often start with the assumption that a reasonable set of options is known beforehand. When this is not the case, there are no definitive answers for which options one should consider. In this paper, we argue that the successor representation (SR), which encodes states based on the pattern of state visitation that follows them, can be seen as a natural substrate for the discovery and use of temporal abstractions. To support our claim, we take a big picture view of recent results, showing how the SR can be used to discover options that facilitate either temporally-extended exploration or planning. We cast these results as instantiations of a general framework for option discovery in which the agent's representation is used to identify useful options, which are then used to further improve its representation. This results in a virtuous, never-ending, cycle in which both the representation and the options are constantly refined based on each other. Beyond option discovery itself, we discuss how the SR allows us to augment a set of options into a combinatorially large counterpart without additional learning. This is achieved through the combination of previously learned options. Our empirical evaluation focuses on options discovered for temporally-extended exploration and on the use of the SR to combine them. The results of our experiments shed light on design decisions involved in the definition of options and demonstrate the synergy of different methods based on the SR, such as eigenoptions and the option keyboard.

We study the model-based reward-free reinforcement learning with linear function approximation for episodic Markov decision processes (MDPs). In this setting, the agent works in two phases. In the exploration phase, the agent interacts with the environment and collects samples without the reward. In the planning phase, the agent is given a specific reward function and uses samples collected from the exploration phase to learn a good policy. We propose a new provably efficient algorithm, called UCRL-RFE under the Linear Mixture MDP assumption, where the transition probability kernel of the MDP can be parameterized by a linear function over certain feature mappings defined on the triplet of state, action, and next state. We show that to obtain an $\epsilon$-optimal policy for arbitrary reward function, UCRL-RFE needs to sample at most $\tilde O(H^5d^2\epsilon^{-2})$ episodes during the exploration phase. Here, $H$ is the length of the episode, $d$ is the dimension of the feature mapping. We also propose a variant of UCRL-RFE using Bernstein-type bonus and show that it needs to sample at most $\tilde O(H^4d(H + d)\epsilon^{-2})$ to achieve an $\epsilon$-optimal policy. By constructing a special class of linear Mixture MDPs, we also prove that for any reward-free algorithm, it needs to sample at least $\tilde \Omega(H^2d\epsilon^{-2})$ episodes to obtain an $\epsilon$-optimal policy. Our upper bound matches the lower bound in terms of the dependence on $\epsilon$ and the dependence on $d$ if $H \ge d$.

Reinforcement learning is a branch in ML, which deals an agent trying to do something in an environment. The agent can be trying to start a fire, when stranded in an island, or the agent can be car trying to park in the right spot. Let's dive in and learn more about reinforcement learning When the agent is trying to do something, it receives a reward when it is getting warmer to what it's supposed to do. The goal of the agent is to try and maximise that reward. The state of an agent is just it's current position with respect to the environment, so if a robot wants to walk, the position of it's legs are it's current state.

Deep reinforcement learning is successfully applied in many real-world robotic tasks. However, it is limited to domains in which a simulator is available or environments that have been tailored and instrumented for the agent's training. Interactive learning approach is useful in training not just industrial robotic systems. Therefore, a recent paper proposes an interactive learning approach in which a human teacher provides evaluative and corrective feedback to the robot during training. The method does not require any reward function and thus avoids credit assignment and reward exploitation issues.

Implicit planning has emerged as an elegant technique for combining learned models of the world with end-to-end model-free reinforcement learning. We study the class of implicit planners inspired by value iteration, an algorithm that is guaranteed to yield perfect policies in fully-specified tabular environments. We find that prior approaches either assume that the environment is provided in such a tabular form -- which is highly restrictive -- or infer "local neighbourhoods" of states to run value iteration over -- for which we discover an algorithmic bottleneck effect. This effect is caused by explicitly running the planning algorithm based on scalar predictions in every state, which can be harmful to data efficiency if such scalars are improperly predicted. We propose eXecuted Latent Value Iteration Networks (XLVINs), which alleviate the above limitations. Our method performs all planning computations in a high-dimensional latent space, breaking the algorithmic bottleneck. It maintains alignment with value iteration by carefully leveraging neural graph-algorithmic reasoning and contrastive self-supervised learning. Across eight low-data settings -- including classical control, navigation and Atari -- XLVINs provide significant improvements to data efficiency against value iteration-based implicit planners, as well as relevant model-free baselines. Lastly, we empirically verify that XLVINs can closely align with value iteration.

Perceived signals in real-world scenarios are usually high-dimensional and noisy, and finding and using their representation that contains essential and sufficient information required by downstream decision-making tasks will help improve computational efficiency and generalization ability in the tasks. In this paper, we focus on partially observable environments and propose to learn a minimal set of state representations that capture sufficient information for decision-making, termed \textit{Action-Sufficient state Representations} (ASRs). We build a generative environment model for the structural relationships among variables in the system and present a principled way to characterize ASRs based on structural constraints and the goal of maximizing cumulative reward in policy learning. We then develop a structured sequential Variational Auto-Encoder to estimate the environment model and extract ASRs. Our empirical results on CarRacing and VizDoom demonstrate a clear advantage of learning and using ASRs for policy learning. Moreover, the estimated environment model and ASRs allow learning behaviors from imagined outcomes in the compact latent space to improve sample efficiency.

Many real-world applications of multi-agent reinforcement learning (RL), such as multi-robot navigation and decentralized control of cyber-physical systems, involve the cooperation of agents as a team with aligned objectives. We study multi-agent RL in the most basic cooperative setting -- Markov teams -- a class of Markov games where the cooperating agents share a common reward. We propose an algorithm in which each agent independently runs stage-based V-learning (a Q-learning style algorithm) to efficiently explore the unknown environment, while using a stochastic gradient descent (SGD) subroutine for policy updates. We show that the agents can learn an $\epsilon$-approximate Nash equilibrium policy in at most $\propto\widetilde{O}(1/\epsilon^4)$ episodes. Our results advocate the use of a novel \emph{stage-based} V-learning approach to create a stage-wise stationary environment. We also show that under certain smoothness assumptions of the team, our algorithm can achieve a nearly \emph{team-optimal} Nash equilibrium. Simulation results corroborate our theoretical findings. One key feature of our algorithm is being \emph{decentralized}, in the sense that each agent has access to only the state and its local actions, and is even \emph{oblivious} to the presence of the other agents. Neither communication among teammates nor coordination by a central controller is required during learning. Hence, our algorithm can readily generalize to an arbitrary number of agents, without suffering from the exponential dependence on the number of agents.

Robots of the future are going to exhibit increasingly human-like and super-human intelligence in a myriad of different tasks. They are also likely going to fail and be incompliant with human preferences in increasingly subtle ways. Towards the goal of achieving autonomous robots, the robot learning community has made rapid strides in applying machine learning techniques to train robots through data and interaction. This makes the study of how best to audit these algorithms for checking their compatibility with humans, pertinent and urgent. In this paper, we draw inspiration from the AI Safety and Alignment communities and make the case that we need to urgently consider ways in which we can best audit our robot learning algorithms to check for failure modes, and ensure that when operating autonomously, they are indeed behaving in ways that the human algorithm designers intend them to. We believe that this is a challenging problem that will require efforts from the entire robot learning community, and do not attempt to provide a concrete framework for auditing. Instead, we outline high-level guidance and a possible approach towards formulating this framework which we hope will serve as a useful starting point for thinking about auditing in the context of robot learning.

Optimizing traffic signal control (TSC) at intersections continues to pose a challenging problem, particularly for large-scale traffic networks. It has been shown in past research that it is feasible to optimize the operations of individual TSC systems or a small number of such systems. However, it has been computationally difficult to scale these solution approaches to large networks partly due to the curse of dimensionality that is encountered as the number of intersections increases. Fortunately, recent studies have recognized the potential of exploiting advancements in deep and reinforcement learning to address this problem, and some preliminary successes have been achieved in this regard. However, facilitating such intelligent solution approaches may require large amounts of infrastructural investments such as roadside units (RSUs) and drones in order to ensure thorough connectivity across all intersections in large networks, an investment that may be burdensome for agencies to undertake. As such, this study builds on recent work to present a scalable TSC model that may reduce the number of required enabling infrastructure. This is achieved using graph attention networks (GATs) to serve as the neural network for deep reinforcement learning, which aids in maintaining the graph topology of the traffic network while disregarding any irrelevant or unnecessary information. A case study is carried out to demonstrate the effectiveness of the proposed model, and the results show much promise. The overall research outcome suggests that by decomposing large networks using fog-nodes, the proposed fog-based graphic RL (FG-RL) model can be easily applied to scale into larger traffic networks.