It remains challenging to deploy existing risk-averse approaches to real-world applications. The reasons are multi-fold, including the lack of global optimality guarantee and the necessity of learning from long-term consecutive trajectories. Long-term consecutive trajectories are prone to involving visiting hazardous states, which is a major concern in the risk-averse setting. This paper proposes Short-Term VOlatility-controlled Policy Search (STOPS), a novel algorithm that solves risk-averse problems by learning from short-term trajectories instead of long-term trajectories. Short-term trajectories are more flexible to generate, and can avoid the danger of hazardous state visitations. By using an actor-critic scheme with an overparameterized two-layer neural network, our algorithm finds a globally optimal policy at a sublinear rate with proximal policy optimization and natural policy gradient, with effectiveness comparable to the state-of-the-art convergence rate of risk-neutral policy-search methods. The algorithm is evaluated on challenging Mujoco robot simulation tasks under the mean-variance evaluation metric. Both theoretical analysis and experimental results demonstrate a state-of-the-art level of STOPS' performance among existing risk-averse policy search methods.

Optimal execution is a sequential decision-making problem for cost-saving in algorithmic trading. Studies have found that reinforcement learning (RL) can help decide the order-splitting sizes. However, a problem remains unsolved: how to place limit orders at appropriate limit prices? The key challenge lies in the "continuous-discrete duality" of the action space. On the one hand, the continuous action space using percentage changes in prices is preferred for generalization. On the other hand, the trader eventually needs to choose limit prices discretely due to the existence of the tick size, which requires specialization for every single stock with different characteristics (e.g., the liquidity and the price range). So we need continuous control for generalization and discrete control for specialization. To this end, we propose a hybrid RL method to combine the advantages of both of them. We first use a continuous control agent to scope an action subset, then deploy a fine-grained agent to choose a specific limit price. Extensive experiments show that our method has higher sample efficiency and better training stability than existing RL algorithms and significantly outperforms previous learning-based methods for order execution.

Multi-goal policy learning for robotic manipulation is challenging. Prior successes have used state-based representations of the objects or provided demonstration data to facilitate learning. In this paper, by hand-coding a high-level discrete representation of the domain, we show that policies to reach dozens of goals can be learned with a single network using Q-learning from pixels. The agent focuses learning on simpler, local policies which are sequenced together by planning in the abstract space. We compare our method against standard multi-goal RL baselines, as well as other methods that leverage the discrete representation, on a challenging block construction domain. We find that our method can build more than a hundred different block structures, and demonstrate forward transfer to structures with novel objects. Lastly, we deploy the policy learned in simulation on a real robot.

Current rapid changes in climate increase the urgency to change energy production and consumption management, to reduce carbon and other green-house gas production. In this context, the French electricity network management company RTE (R{\'e}seau de Transport d'{\'E}lectricit{\'e}) has recently published the results of an extensive study outlining various scenarios for tomorrow's French power management. We propose a challenge that will test the viability of such a scenario. The goal is to control electricity transportation in power networks, while pursuing multiple objectives: balancing production and consumption, minimizing energetic losses, and keeping people and equipment safe and particularly avoiding catastrophic failures. While the importance of the application provides a goal in itself, this challenge also aims to push the state-of-the-art in a branch of Artificial Intelligence (AI) called Reinforcement Learning (RL), which offers new possibilities to tackle control problems. In particular, various aspects of the combination of Deep Learning and RL called Deep Reinforcement Learning remain to be harnessed in this application domain. This challenge belongs to a series started in 2019 under the name "Learning to run a power network" (L2RPN). In this new edition, we introduce new more realistic scenarios proposed by RTE to reach carbon neutrality by 2050, retiring fossil fuel electricity production, increasing proportions of renewable and nuclear energy and introducing batteries. Furthermore, we provide a baseline using state-of-the-art reinforcement learning algorithm to stimulate the future participants.

Closed-loop reservoir management (CLRM), in which history matching and production optimization are performed multiple times over the life of an asset, can provide significant improvement in the specified objective. These procedures are computationally expensive due to the large number of flow simulations required for data assimilation and optimization. Existing CLRM procedures are applied asset by asset, without utilizing information that could be useful over a range assets. Here, we develop a CLRM framework for multiple assets with varying numbers of wells. We use deep reinforcement learning to train a single global control policy that is applicable for all assets considered. The new framework is an extension of a recently introduced control policy methodology for individual assets. Embedding layers are incorporated into the representation to handle the different numbers of decision variables that arise for the different assets. Because the global control policy learns a unified representation of useful features from multiple assets, it is less expensive to construct than asset-by-asset training (we observe about 3x speedup in our examples). The production optimization problem includes a relative-change constraint on the well settings, which renders the results suitable for practical use. We apply the multi-asset CLRM framework to 2D and 3D water-flooding examples. In both cases, four assets with different well counts, well configurations, and geostatistical descriptions are considered. Numerical experiments demonstrate that the global control policy provides objective function values, for both the 2D and 3D cases, that are nearly identical to those from control policies trained individually for each asset. This promising finding suggests that multi-asset CLRM may indeed represent a viable practical strategy.

Warehouse Management Systems have been evolving and improving thanks to new Data Intelligence techniques. However, many current optimizations have been applied to specific cases or are in great need of manual interaction. Here is where Reinforcement Learning techniques come into play, providing automatization and adaptability to current optimization policies. In this paper, we present Storehouse, a customizable environment that generalizes the definition of warehouse simulations for Reinforcement Learning. We also validate this environment against state-of-the-art reinforcement learning algorithms and compare these results to human and random policies.

Clinically significant prostate cancer has a better chance to be sampled during ultrasound-guided biopsy procedures, if suspected lesions found in pre-operative magnetic resonance (MR) images are used as targets. However, the diagnostic accuracy of the biopsy procedure is limited by the operator-dependent skills and experience in sampling the targets, a sequential decision making process that involves navigating an ultrasound probe and placing a series of sampling needles for potentially multiple targets. This work aims to learn a reinforcement learning (RL) policy that optimises the actions of continuous positioning of 2D ultrasound views and biopsy needles with respect to a guiding template, such that the MR targets can be sampled efficiently and sufficiently. We first formulate the task as a Markov decision process (MDP) and construct an environment that allows the targeting actions to be performed virtually for individual patients, based on their anatomy and lesions derived from MR images. A patient-specific policy can thus be optimised, before each biopsy procedure, by rewarding positive sampling in the MDP environment. Experiment results from fifty four prostate cancer patients show that the proposed RL-learned policies obtained a mean hit rate of 93% and an average cancer core length of 11 mm, which compared favourably to two alternative baseline strategies designed by humans, without hand-engineered rewards that directly maximise these clinically relevant metrics. Perhaps more interestingly, it is found that the RL agents learned strategies that were adaptive to the lesion size, where spread of the needles was prioritised for smaller lesions. Such a strategy has not been previously reported or commonly adopted in clinical practice, but led to an overall superior targeting performance, achieving higher hit rates (93% vs 76%) and measured cancer core lengths (11.0mm vs 9.8mm) when compared with intuitively designed strategies.

Despite the success of AI-enabled onboard perception, on-ramp merging has been one of the main challenges for autonomous driving. Due to limited sensing range of onboard sensors, a merging vehicle can hardly observe main road conditions and merge properly. By leveraging the wireless communications between connected and automated vehicles (CAVs), a merging CAV has potential to proactively obtain the intentions of nearby vehicles. However, CAVs can be prone to inaccurate observations, such as the noisy basic safety messages (BSM) and poor quality surveillance images. In this paper, we present a novel approach for Robust on-ramp merge of CAVs via Augmented and Multi-modal Reinforcement Learning, named by RAMRL. Specifically, we formulate the on-ramp merging problem as a Markov decision process (MDP) by taking driving safety, comfort driving behavior, and traffic efficiency into account. To provide reliable merging maneuvers, we simultaneously leverage BSM and surveillance images for multi-modal observation, which is used to learn a policy model through proximal policy optimization (PPO). Moreover, to improve data efficiency and provide better generalization performance, we train the policy model with augmented data (e.g., noisy BSM and noisy surveillance images). Extensive experiments are conducted with Simulation of Urban MObility (SUMO) platform under two typical merging scenarios. Experimental results demonstrate the effectiveness and efficiency of our robust on-ramp merging design.

The capacity sharing problem in Radio Access Network (RAN) slicing deals with the distribution of the capacity available in each RAN node among various RAN slices to satisfy their traffic demands and efficiently use the radio resources. While several capacity sharing algorithmic solutions have been proposed in the literature, their practical implementation still remains as a gap. In this paper, the implementation of a Reinforcement Learning-based capacity sharing algorithm over the O-RAN architecture is discussed, providing insights into the operation of the involved interfaces and the containerization of the solution. Moreover, the description of the testbed implemented to validate the solution is included and some performance and validation results are presented.

Impressive results in natural language processing (NLP) based on the Transformer neural network architecture have inspired researchers to explore viewing offline reinforcement learning (RL) as a generic sequence modeling problem. Recent works based on this paradigm have achieved state-of-the-art results in several of the mostly deterministic offline Atari and D4RL benchmarks. However, because these methods jointly model the states and actions as a single sequencing problem, they struggle to disentangle the effects of the policy and world dynamics on the return. Thus, in adversarial or stochastic environments, these methods lead to overly optimistic behavior that can be dangerous in safety-critical systems like autonomous driving. In this work, we propose a method that addresses this optimism bias by explicitly disentangling the policy and world models, which allows us at test time to search for policies that are robust to multiple possible futures in the environment. We demonstrate our method's superior performance on a variety of autonomous driving tasks in simulation.