Generative flow networks (GFlowNets), as an emerging technique, can be used as an alternative to reinforcement learning for exploratory control tasks. GFlowNet aims to generate distribution proportional to the rewards over terminating states, and to sample different candidates in an active learning fashion. GFlowNets need to form a DAG and compute the flow matching loss by traversing the inflows and outflows of each node in the trajectory. No experiments have yet concluded that GFlowNets can be used to handle continuous tasks. In this paper, we propose generative continuous flow networks (CFlowNets) that can be applied to continuous control tasks. First, we present the theoretical formulation of CFlowNets. Then, a training framework for CFlowNets is proposed, including the action selection process, the flow approximation algorithm, and the continuous flow matching loss function. Afterward, we theoretically prove the error bound of the flow approximation. The error decreases rapidly as the number of flow samples increases. Finally, experimental results on continuous control tasks demonstrate the performance advantages of CFlowNets compared to many reinforcement learning methods, especially regarding exploration ability.

Uncertainty quantification has been extensively used as a means to achieve efficient directed exploration in Reinforcement Learning (RL). However, state-of-the-art methods for continuous actions still suffer from high sample complexity requirements. Indeed, they either completely lack strategies for propagating the epistemic uncertainty throughout the updates, or they mix it with aleatoric uncertainty while learning the full return distribution (e.g., distributional RL). In this paper, we propose Wasserstein Actor-Critic (WAC), an actor-critic architecture inspired by the recent Wasserstein Q-Learning (WQL) \citep{wql}, that employs approximate Q-posteriors to represent the epistemic uncertainty and Wasserstein barycenters for uncertainty propagation across the state-action space. WAC enforces exploration in a principled way by guiding the policy learning process with the optimization of an upper bound of the Q-value estimates. Furthermore, we study some peculiar issues that arise when using function approximation, coupled with the uncertainty estimation, and propose a regularized loss for the uncertainty estimation. Finally, we evaluate our algorithm on standard MujoCo tasks as well as suite of continuous-actions domains, where exploration is crucial, in comparison with state-of-the-art baselines.

Reinforcement learning (RL) has recently proven great success in various domains. Yet, the design of the reward function requires detailed domain expertise and tedious fine-tuning to ensure that agents are able to learn the desired behaviour. Using a sparse reward conveniently mitigates these challenges. However, the sparse reward represents a challenge on its own, often resulting in unsuccessful training of the agent. In this paper, we therefore address the sparse reward problem in RL. Our goal is to find an effective alternative to reward shaping, without using costly human demonstrations, that would also be applicable to a wide range of domains. Hence, we propose to use model predictive control~(MPC) as an experience source for training RL agents in sparse reward environments. Without the need for reward shaping, we successfully apply our approach in the field of mobile robot navigation both in simulation and real-world experiments with a Kuboki Turtlebot 2. We furthermore demonstrate great improvement over pure RL algorithms in terms of success rate as well as number of collisions and timeouts. Our experiments show that MPC as an experience source improves the agent's learning process for a given task in the case of sparse rewards.

The multi-agent setting is intricate and unpredictable since the behaviors of multiple agents influence one another. To address this environmental uncertainty, distributional reinforcement learning algorithms that incorporate uncertainty via distributional output have been integrated with multi-agent reinforcement learning (MARL) methods, achieving state-of-the-art performance. However, distributional MARL algorithms still rely on the traditional $\epsilon$-greedy, which does not take cooperative strategy into account. In this paper, we present a risk-based exploration that leads to collaboratively optimistic behavior by shifting the sampling region of distribution. Initially, we take expectations from the upper quantiles of state-action values for exploration, which are optimistic actions, and gradually shift the sampling region of quantiles to the full distribution for exploitation. By ensuring that each agent is exposed to the same level of risk, we can force them to take cooperatively optimistic actions. Our method shows remarkable performance in multi-agent settings requiring cooperative exploration based on quantile regression appropriately controlling the level of risk.

Reinforcement Learning (RL) is an area of machine learning figuring out how agents take actions in an unknown environment to maximize its rewards. Unlike classical Markov Decision Process (MDP) in which agent has full knowledge of its state, rewards, and transitional probability, reinforcement learning utilizes exploration and exploitation for the model uncertainty. Under the condition that the model usually has a large state space, a neural network (NN) can be used to correlate its input state to its output actions to maximize the agent's rewards. However, building and training an efficient neural network is challenging. Inspired by Double Q-learning and Asynchronous Advantage Actor-Critic (A3C) algorithm, we will propose and implement an improved version of Double A3C algorithm which utilizing the strength of both algorithms to play OpenAI Gym Atari 2600 games to beat its benchmarks for our project.

Altitude-controlled balloons hold great promise for performing high-priority scientific investigations of Venus's atmosphere and geological phenomena, including tectonic and volcanic activity, as demonstrated by a number of recent Earth-based experiments. In this paper, we explore a concept of operations where multiple autonomous, altitude-controlled balloons monitor explosive volcanic activity on Venus through infrasound microbarometers, and autonomously navigate the uncertain wind field to perform follow-on observations of detected events of interest. We propose a novel autonomous guidance technique for altitude-controlled balloons in Venus's uncertain wind field, and show the approach can result in an increase of up to 63% in the number of close-up observations of volcanic events compared to passive drifters, and a 16% increase compared to ground-in-the-loop guidance. The results are robust to uncertainty in the wind field, and hold across large changes in the frequency of explosive volcanic events, sensitivity of the microbarometer detectors, and numbers of aerial platforms.

Existing reinforcement learning environment libraries use monolithic environment classes, provide shallow methods for altering agent observation and action spaces, and/or are tied to a specific simulation environment. The Core Reinforcement Learning library (CoRL) is a modular, composable, and hyper-configurable environment creation tool. It allows minute control over agent observations, rewards, and done conditions through the use of easy-to-read configuration files, pydantic validators, and a functor design pattern. Using integration pathways allows agents to be quickly implemented in new simulation environments, encourages rapid exploration, and enables transition of knowledge from low-fidelity to high-fidelity simulations. Natively multi-agent design and integration with Ray/RLLib (Liang et al., 2018) at release allow for easy scalability of agent complexity and computing power. The code is publicly released and available at https://github.com/act3-ace/CoRL.

Controlling agents remotely with deep reinforcement learning~(DRL) in the real world is yet to come. One crucial stepping stone is to devise RL algorithms that are robust in the face of dropped information from corrupted communication or malfunctioning sensors. Typical RL methods usually require considerable online interaction data that are costly and unsafe to collect in the real world. Furthermore, when applying to the frame dropping scenarios, they perform unsatisfactorily even with moderate drop rates. To address these issues, we propose Decision Transformer under Random Frame Dropping~(DeFog), an offline RL algorithm that enables agents to act robustly in frame dropping scenarios without online interaction. DeFog first randomly masks out data in the offline datasets and explicitly adds the time span of frame dropping as inputs. After that, a finetuning stage on the same offline dataset with a higher mask rate would further boost the performance. Empirical results show that DeFog outperforms strong baselines under severe frame drop rates like 90\%, while maintaining similar returns under non-frame-dropping conditions in the regular MuJoCo control benchmarks and the Atari environments. Our approach offers a robust and deployable solution for controlling agents in real-world environments with limited or unreliable data.

Standard Markov decision process (MDP) and reinforcement learning algorithms optimize the policy with respect to the expected gain. We propose an algorithm which enables to optimize an alternative objective: the probability that the gain is greater than a given value. The algorithm can be seen as an extension of the value iteration algorithm. We also show how the proposed algorithm could be generalized to use neural networks, similarly to the deep Q learning extension of Q learning.

Real world applications of Reinforcement Learning (RL) are often partially observable, thus requiring memory. Despite this, partial observability is still largely ignored by contemporary RL benchmarks and libraries. We introduce Partially Observable Process Gym (POPGym), a two-part library containing (1) a diverse collection of 15 partially observable environments, each with multiple difficulties and (2) implementations of 13 memory model baselines -- the most in a single RL library. Existing partially observable benchmarks tend to fixate on 3D visual navigation, which is computationally expensive and only one type of POMDP. In contrast, POPGym environments are diverse, produce smaller observations, use less memory, and often converge within two hours of training on a consumer-grade GPU. We implement our high-level memory API and memory baselines on top of the popular RLlib framework, providing plug-and-play compatibility with various training algorithms, exploration strategies, and distributed training paradigms. Using POPGym, we execute the largest comparison across RL memory models to date. POPGym is available at https://github.com/proroklab/popgym.