The AlphaZero/MuZero (A/MZ) family of algorithms has achieved remarkable success across various challenging domains by integrating Monte Carlo Tree Search (MCTS) with learned models. Learned models introduce epistemic uncertainty, which is caused by learning from limited data and is useful for exploration in sparse reward environments. MCTS does not account for the propagation of this uncertainty however. To address this, we introduce Epistemic MCTS (EMCTS): a theoretically motivated approach to account for the epistemic uncertainty in search and harness the search for deep exploration. In the challenging sparse-reward task of writing code in the Assembly language subleq, AZ paired with our method achieves significantly higher sample efficiency over baseline AZ. Search with EMCTS solves variations of the commonly used hard-exploration benchmark Deep Sea - which baseline A/MZ are practically unable to solve - much faster than an otherwise equivalent method that does not use search for uncertainty estimation, demonstrating significant benefits from search for epistemic uncertainty estimation.

In contrast to classical reinforcement learning, distributional RL algorithms aim to learn the distribution of returns rather than their expected value. Since the nature of the return distribution is generally unknown a priori or arbitrarily complex, a common approach finds approximations within a set of representable, parametric distributions. Typically, this involves a projection of the unconstrained distribution onto the set of simplified distributions. We argue that this projection step entails a strong inductive bias when coupled with neural networks and gradient descent, thereby profoundly impacting the generalization behavior of learned models. In order to facilitate reliable uncertainty estimation through diversity, this work studies the combination of several different projections and representations in a distributional ensemble. We establish theoretical properties of such projection ensembles and derive an algorithm that uses ensemble disagreement, measured by the average
-Wasserstein distance, as a bonus for deep exploration. We evaluate our algorithm on the behavior suite benchmark and find that diverse projection ensembles lead to significant performance improvements over existing methods on a wide variety of tasks with the most pronounced gains in directed exploration problems.

Smart cameras are an essential component in surveillance and monitoring applications, and they have been typically deployed in networks of fixed camera locations. The addition of mobile cameras, mounted on robots, can overcome some of the limitations of static networks such as blind spots or back-lightning, allowing the system to gather the best information at each time by active positioning. This work presents a hybrid camera system, with static and mobile cameras, where all the cameras collaborate to observe people moving freely in the environment and efficiently visualize certain attributes from each person. Our solution combines a multi-camera distributed tracking system, to localize with precision all the people, with a control scheme that moves the mobile cameras to the best viewpoints for a specific classification task. The main contribution of this paper is a novel framework that exploits the synergies that result from the cooperation of the tracking and the control modules, obtaining a system closer to the real-world application and capable of high-level scene understanding. The static camera network provides global awareness of the control scheme to move the robots. In exchange, the mobile cameras onboard the robots provide enhanced information about the people on the scene. We perform a thorough analysis of the people monitoring application performance under different conditions thanks to the use of a photo-realistic simulation environment. Our experiments demonstrate the benefits of collaborative mobile cameras with respect to static or individual camera setups.

In reinforcement learning (RL), different reward functions can define the same optimal policy but result in drastically different learning performance. For some, the agent gets stuck with a suboptimal behavior, and for others, it solves the task efficiently. Choosing a good reward function is hence an extremely important yet challenging problem. In this paper, we explore an alternative approach for using rewards for learning. We introduce max-reward RL, where an agent optimizes the maximum rather than the cumulative reward. Unlike earlier works, our approach works for deterministic and stochastic environments and can be easily combined with state-of-the-art RL algorithms. In the experiments, we study the performance of max-reward RL algorithms in two goal-reaching environments from Gymnasium-Robotics and demonstrate its benefits over standard RL. The code is available at https://github.com/veviurko/To-the-Max.

In this paper, we address the problem of real-time motion planning for multiple robotic manipulators that operate in close proximity. We build upon the concept of dynamic fabrics and extend them to multi-robot systems, referred to as Multi-Robot Dynamic Fabrics (MRDF). This geometric method enables a very high planning frequency for high-dimensional systems at the expense of being reactive and prone to deadlocks. To detect and resolve deadlocks, we propose Rollout Fabrics where MRDF are forward simulated in a decentralized manner. We validate the methods in simulated close-proximity pick-and-place scenarios with multiple manipulators, showing high-success rates and real-time performance. Code, video: https://github.com/tud-amr/multi-robot-fabrics

Many modern reinforcement learning algorithms build on the actor-critic (AC) framework: iterative improvement of a policy (the actor) using policy improvement operators and iterative approximation of the policy's value (the critic). In contrast, the popular value-based algorithm family employs improvement operators in the value update, to iteratively improve the value function directly. In this work, we propose a general extension to the AC framework that employs two separate improvement operators: one applied to the policy in the spirit of policy-based algorithms and one applied to the value in the spirit of value-based algorithms, which we dub Value-Improved AC (VI-AC). We design two practical VI-AC algorithms based in the popular online off-policy AC algorithms TD3 and DDPG. We evaluate VI-TD3 and VI-DDPG in the Mujoco benchmark and find that both improve upon or match the performance of their respective baselines in all environments tested

One of the most well-studied and highly performing planning approaches used in Model-Based Reinforcement Learning (MBRL) is Monte-Carlo Tree Search (MCTS). Key challenges of MCTS-based MBRL methods remain dedicated deep exploration and reliability in the face of the unknown, and both challenges can be alleviated through principled epistemic uncertainty estimation in the predictions of MCTS. We present two main contributions: First, we develop methodology to propagate epistemic uncertainty in MCTS, enabling agents to estimate the epistemic uncertainty in their predictions. Second, we utilize the propagated uncertainty for a novel deep exploration algorithm by explicitly planning to explore. We incorporate our approach into variations of MCTS-based MBRL approaches with learned and provided dynamics models, and empirically show deep exploration through successful epistemic uncertainty estimation achieved by our approach. We compare to a non-planning-based deep-exploration baseline, and demonstrate that planning with epistemic MCTS significantly outperforms non-planning based exploration in the investigated deep exploration benchmark.

In reinforcement learning (RL), key components of many algorithms are the exploration strategy and replay buffer. These strategies regulate what environment data is collected and trained on and have been extensively studied in the RL literature. In this paper, we investigate the impact of these components in the context of generalisation in multi-task RL. We investigate the hypothesis that collecting and training on more diverse data from the training environments will improve zero-shot generalisation to new tasks. We motivate mathematically and show empirically that generalisation to tasks that are "reachable'' during training is improved by increasing the diversity of transitions in the replay buffer. Furthermore, we show empirically that this same strategy also shows improvement for generalisation to similar but "unreachable'' tasks which could be due to improved generalisation of the learned latent representations.

Decentralized multi-robot systems typically perform coordinated motion planning by constantly broadcasting their intentions to avoid collisions. However, the risk of collision between robots varies as they move and communication may not always be needed. This paper presents an efficient communication method that addresses the problem of “when” and “with whom” to communicate in multi-robot collision avoidance scenarios. In this approach, each robot learns to reason about other robots’ states and considers the risk of future collisions before asking for the trajectory plans of other robots. We introduce a new neural architecture for the learned communication policy which allows our method to be scalable. We evaluate and verify the proposed communication strategy in simulation with up to twelve quadrotors, and present results on the zero-shot generalization/robustness capabilities of the policy in different scenarios. We demonstrate that our policy (learned in a simulated environment) can be successfully transferred to real robots.

In contrast to classical reinforcement learning, distributional reinforcement learning algorithms aim to learn the distribution of returns rather than their expected value. Since the nature of the return distribution is generally unknown a priori or arbitrarily complex, a common approach finds approximations within a set of representable, parametric distributions. Typically, this involves a projection of the unconstrained distribution onto the set of simplified distributions. We argue that this projection step entails a strong inductive bias when coupled with neural networks and gradient descent, thereby profoundly impacting the generalization behavior of learned models. In order to facilitate reliable uncertainty estimation through diversity, this work studies the combination of several different projections and representations in a distributional ensemble. We establish theoretical properties of such projection ensembles and derive an algorithm that uses ensemble disagreement, measured by the average 1-Wasserstein distance, as a bonus for deep exploration. We evaluate our algorithm on the behavior suite benchmark and find that diverse projection ensembles lead to significant performance improvements over existing methods on a wide variety of tasks with the most pronounced gains in directed exploration problems.

In this paper, we consider the problem where a drone has to collect semantic information to classify multiple moving targets. In particular, we address the challenge of computing control inputs that move the drone to informative viewpoints, position and orientation, when the information is extracted using a “black-box” classifier, e.g., a deep learning neural network. These algorithms typically lack of analytical relationships between the viewpoints and their associated outputs, preventing their use in information-gathering schemes. To fill this gap, we propose a novel attention-based architecture, trained via Reinforcement Learning (RL), that outputs the next viewpoint for the drone favoring the acquisition of evidence from as many unclassified targets as possible while reasoning about their movement, orientation, and occlusions. Then, we use a low-level MPC controller to move the drone to the desired viewpoint taking into account its actual dynamics. We show that our approach not only outperforms a variety of baselines but also generalizes to scenarios unseen during training. Additionally, we show that the network scales to large numbers of targets and generalizes well to different movement dynamics of the targets.

Many electric vehicles (EVs) are using today’s distribution grids, and their flexibility can be highly beneficial for the grid operators. This flexibility can be best exploited by DC power networks, as they allow charging and discharging without extra power electronics and transformation losses. From the grid control perspective, algorithms for planning EV charging are necessary. This paper studies the problem of EV charging planning under limited grid capacity and extends it to the partially observable case. We demonstrate how limited information about the EV locations in a grid may disrupt the operation planning in DC grids with tight constraints. We introduce two methods to change the grid topology such that partial observability of the EV locations is resolved. The suggested models are evaluated on the IEEE 16 bus system and multiple randomly generated grids with varying capacities. The experiments show that these methods efficiently solve the partially observable EV charging planning problem and offer a trade-off between computational time and performance.

VDN and QMIX are two popular value-based algorithms for cooperative MARL that learn a centralized action value function as a monotonic mixing of per-agent utilities. While this enables easy decentralization of the learned policy, the restricted joint action value function can prevent them from solving tasks that require significant coordination between agents at a given timestep. We show that this problem can be overcome by improving the joint exploration of all agents during training. Specifically, we propose a novel MARL approach called Universal Value Exploration (UneVEn) that learns a set of related tasks simultaneously with a linear decomposition of universal successor features. With the policies of already solved related tasks, the joint exploration process of all agents can be improved to help them achieve better coordination. Empirical results on a set of exploration games, challenging cooperative predator-prey tasks requiring significant coordination among agents, and StarCraft II micromanagement benchmarks show that UneVEn can solve tasks where other state-of-the-art MARL methods fail.

Multitask Reinforcement Learning is a promising way to obtain models with better performance, generalisation, data efficiency, and robustness. Most existing work is limited to compatible settings, where the state and action space dimensions are the same across tasks. Graph Neural Networks (GNN) are one way to address incompatible environments, because they can process graphs of arbitrary size. They also allow practitioners to inject biases encoded in the structure of the input graph. Existing work in graph-based continuous control uses the physical morphology of the agent to construct the input graph, i.e., encoding limb features as node labels and using edges to connect the nodes if their corresponded limbs are physically connected. In this work, we present a series of ablations on existing methods that show that morphological information encoded in the graph does not improve their performance. Motivated by the hypothesis that any benefits GNNs extract from the graph structure are outweighed by difficulties they create for message passing, we also propose Amorpheus, a transformer-based approach. Further results show that, while Amorpheus ignores the morphological information that GNNs encode, it nonetheless substantially outperforms GNN-based methods.

Combinatorial optimization (CO) problems are at the heart of both practical and theoretical research. Due to their complexity, many problems cannot be solved via exact methods in reasonable time; hence, we resort to heuristic solution methods. In recent years, machine learning (ML) has brought immense benefits in many research areas, including heuristic solution methods for CO problems. Among ML methods, reinforcement learning (RL) seems to be the most promising method to find good solutions for CO problems. In this work, we investigate an RL framework, whose agent is based on self-attention, to achieve solutions for the knapsack problem, which is a CO problem. Our algorithm finds close to optimal solutions for instances up to one hundred items, which leads to conjecture that RL and self-attention may be major building blocks for future state-of-the-art heuristics for other CO problems.

Real world multi-agent tasks often involve varying types and quantities of agents and non-agent entities; however, agents within these tasks rarely need to consider all others at all times in order to act effectively. Factored value function approaches have historically leveraged such independences to improve learning efficiency, but these approaches typically rely on domain knowledge to select fixed subsets of state features to include in each factor. We propose to utilize value function factoring with random subsets of entities in each factor as an auxiliary objective in order to disentangle value predictions from irrelevant entities. This factoring approach is instantiated through a simple attention mechanism masking procedure. We hypothesize that such an approach helps agents learn more effectively in multi-agent settings by discovering common trajectories across episodes within sub-groups of agents/entities. Our approach, Randomized Entity-wise Factorization for Imagined Learning (REFIL), outperforms all strong baselines by a significant margin in challenging StarCraft micromanagement tasks.

Non-stationarity can arise in Reinforcement Learning (RL) even in stationary environments. For example, most RL algorithms collect new data throughout training, using a non-stationary behaviour policy. Due to the transience of this non-stationarity, it is often not explicitly addressed in deep RL and a single neural network is continually updated. However, we find evidence that neural networks exhibit a memory effect, where these transient non-stationarities can permanently impact the latent representation and adversely affect generalisation performance. Consequently, to improve generalisation of deep RL agents, we propose Iterated Relearning (ITER). ITER augments standard RL training by repeated knowledge transfer of the current policy into a freshly initialised network, which thereby experiences less non-stationarity during training. Experimentally, we show that ITER improves performance on the challenging generalisation benchmarks ProcGen and Multiroom.

We propose FACtored Multi-Agent Centralised policy gradients (FACMAC), a new method for cooperative multi-agent reinforcement learning in both discrete and continuous action spaces. Like MADDPG, a popular multi-agent actor-critic method, our approach uses deep deterministic policy gradients to learn policies. However, FACMAC learns a centralised but factored critic, which combines per-agent utilities into the joint action-value function via a non-linear monotonic function, as in QMIX, a popular multi-agent Q-learning algorithm. However, unlike QMIX, there are no inherent constraints on factoring the critic. We thus also employ a nonmonotonic factorisation and empirically demonstrate that its increased representational capacity allows it to solve some tasks that cannot be solved with monolithic, or monotonically factored critics. In addition, FACMAC uses a centralised policy gradient estimator that optimises over the entire joint action space, rather than optimising over each agent’s action space separately as in MADDPG. This allows for more coordinated policy changes and fully reaps the benefits of a centralised critic. We evaluate FACMAC on variants of the multi-agent particle environments, a novel multi-agent MuJoCo benchmark, and a challenging set of StarCraft II micromanagement tasks. Empirical results demonstrate FACMAC’s superior performance over MADDPG and other baselines on all three domains.

This paper introduces the deep coordination graph (DCG) for collaborative multi-agent reinforcement learning. DCG strikes a flexible tradeoff between representational capacity and generalization by factoring the joint value function of all agents according to a coordination graph into payoffs between pairs of agents. The value can be maximized by local message passing along the graph, which allows training of the value function end-to-end with Q-learning. Payoff functions are approximated with deep neural networks that employ parameter sharing and low-rank approximations to significantly improve sample efficiency. We show that DCG can solve predatorprey tasks that highlight the relative overgeneralization pathology, as well as challenging StarCraft II micromanagement tasks.