This work contributes to developing an agent based on deep reinforcement learning capable of acting in a beyond visual range (BVR) air combat simulation environment. The paper presents an overview of building an agent representing a high-performance fighter aircraft that can learn and improve its role in BVR combat over time based on rewards calculated using operational metrics. Also, through self-play experiments, it expects to generate new air combat tactics never seen before. Finally, we hope to examine a real pilot's ability, using virtual simulation, to interact in the same environment with the trained agent and compare their performances. This research will contribute to the air combat training context by developing agents that can interact with real pilots to improve their performances in air defense missions.

This work investigates the formal policy synthesis of continuous-state stochastic dynamic systems given high-level specifications in linear temporal logic. To learn an optimal policy that maximizes the satisfaction probability, we take a product between a dynamic system and the translated automaton to construct a product system on which we solve an optimal planning problem. Since this product system has a hybrid product state space that results in reward sparsity, we introduce a generalized optimal backup order, in reverse to the topological order, to guide the value backups and accelerate the learning process. We provide the optimality proof for using the generalized optimal backup order in this optimal planning problem. Further, this paper presents an actor-critic reinforcement learning algorithm when topological order applies. This algorithm leverages advanced mathematical techniques and enjoys the property of hyperparameter self-tuning. We provide proof of the optimality and convergence of our proposed reinforcement learning algorithm. We use neural networks to approximate the value function and policy function for hybrid product state space. Furthermore, we observe that assigning integer numbers to automaton states can rank the value or policy function approximated by neural networks. To break the ordinal relationship, we use an individual neural network for each automaton state's value (policy) function, termed modular learning. We conduct two experiments. First, to show the efficacy of our reinforcement learning algorithm, we compare it with baselines on a classic control task, CartPole. Second, we demonstrate the empirical performance of our formal policy synthesis framework on motion planning of a Dubins car with a temporal specification.

We present a system that enables an autonomous small-scale RC car to drive aggressively from visual observations using reinforcement learning (RL). Our system, FastRLAP (faster lap), trains autonomously in the real world, without human interventions, and without requiring any simulation or expert demonstrations. Our system integrates a number of important components to make this possible: we initialize the representations for the RL policy and value function from a large prior dataset of other robots navigating in other environments (at low speed), which provides a navigation-relevant representation. From here, a sample-efficient online RL method uses a single low-speed user-provided demonstration to determine the desired driving course, extracts a set of navigational checkpoints, and autonomously practices driving through these checkpoints, resetting automatically on collision or failure. Perhaps surprisingly, we find that with appropriate initialization and choice of algorithm, our system can learn to drive over a variety of racing courses with less than 20 minutes of online training. The resulting policies exhibit emergent aggressive driving skills, such as timing braking and acceleration around turns and avoiding areas which impede the robot's motion, approaching the performance of a human driver using a similar first-person interface over the course of training.

Attitude control of fixed-wing unmanned aerial vehicles (UAVs) is a difficult control problem in part due to uncertain nonlinear dynamics, actuator constraints, and coupled longitudinal and lateral motions. Current state-of-the-art autopilots are based on linear control and are thus limited in their effectiveness and performance. Deep reinforcement learning (DRL) is a machine learning method to automatically discover optimal control laws through interaction with the controlled system, which can handle complex nonlinear dynamics. We show in this paper that DRL can successfully learn to perform attitude control of a fixed-wing UAV operating directly on the original nonlinear dynamics, requiring as little as three minutes of flight data. We initially train our model in a simulation environment and then deploy the learned controller on the UAV in flight tests, demonstrating comparable performance to the state-of-the-art ArduPlane proportional-integral-derivative (PID) attitude controller with no further online learning required. Learning with significant actuation delay and diversified simulated dynamics were found to be crucial for successful transfer to control of the real UAV. In addition to a qualitative comparison with the ArduPlane autopilot, we present a quantitative assessment based on linear analysis to better understand the learning controller's behavior.

Legged robots have enormous potential in their range of capabilities, from navigating unstructured terrains to high-speed running. However, designing robust controllers for highly agile dynamic motions remains a substantial challenge for roboticists. Reinforcement learning (RL) offers a promising data-driven approach for automatically training such controllers. However, exploration in these high-dimensional, underactuated systems remains a significant hurdle for enabling legged robots to learn performant, naturalistic, and versatile agility skills. We propose a framework for training complex robotic skills by transferring experience from existing controllers to jumpstart learning new tasks. To leverage controllers we can acquire in practice, we design this framework to be flexible in terms of their source -- that is, the controllers may have been optimized for a different objective under different dynamics, or may require different knowledge of the surroundings -- and thus may be highly suboptimal for the target task. We show that our method enables learning complex agile jumping behaviors, navigating to goal locations while walking on hind legs, and adapting to new environments. We also demonstrate that the agile behaviors learned in this way are graceful and safe enough to deploy in the real world.

In this paper, we review the question of which action space is best suited for controlling a real biped robot in combination with Sim2Real training. Position control has been popular as it has been shown to be more sample efficient and intuitive to combine with other planning algorithms. However, for position control gain tuning is required to achieve the best possible policy performance. We show that instead, using a torque-based action space enables task-and-robot agnostic learning with less parameter tuning and mitigates the sim-to-reality gap by taking advantage of torque control's inherent compliance. Also, we accelerate the torque-based-policy training process by pre-training the policy to remain upright by compensating for gravity. The paper showcases the first successful sim-to-real transfer of a torque-based deep reinforcement learning policy on a real human-sized biped robot. The video is available at https://youtu.be/CR6pTS39VRE.

Route planning is essential to mobile robot navigation problems. In recent years, deep reinforcement learning (DRL) has been applied to learning optimal planning policies in stochastic environments without prior knowledge. However, existing works focus on learning policies that maximize the expected return, the performance of which can vary greatly when the level of stochasticity in the environment is high. In this work, we propose a distributional reinforcement learning based framework that learns return distributions which explicitly reflect environmental stochasticity. Policies based on the second-order stochastic dominance (SSD) relation can be used to make adjustable route decisions according to user preference on performance robustness. Our proposed method is evaluated in a simulated road network environment, and experimental results show that our method is able to plan the shortest routes that minimize stochasticity in travel time when robustness is preferred, while other state-of-the-art DRL methods are agnostic to environmental stochasticity.

This paper considers the problem of learning safe policies in the context of reinforcement learning (RL). In particular, we consider the notion of probabilistic safety. This is, we aim to design policies that maintain the state of the system in a safe set with high probability. This notion differs from cumulative constraints often considered in the literature. The challenge of working with probabilistic safety is the lack of expressions for their gradients. Indeed, policy optimization algorithms rely on gradients of the objective function and the constraints. To the best of our knowledge, this work is the first one providing such explicit gradient expressions for probabilistic constraints. It is worth noting that the gradient of this family of constraints can be applied to various policy-based algorithms. We demonstrate empirically that it is possible to handle probabilistic constraints in a continuous navigation problem.

With the widespread adoption of deep learning, reinforcement learning (RL) has experienced a dramatic increase in popularity, scaling to previously intractable problems, such as playing complex games from pixel observations, sustaining conversations with humans, and controlling robotic agents. However, there is still a wide range of domains inaccessible to RL due to the high cost and danger of interacting with the environment. Offline RL is a paradigm that learns exclusively from static datasets of previously collected interactions, making it feasible to extract policies from large and diverse training datasets. Effective offline RL algorithms have a much wider range of applications than online RL, being particularly appealing for real-world applications, such as education, healthcare, and robotics. In this work, we contribute with a unifying taxonomy to classify offline RL methods. Furthermore, we provide a comprehensive review of the latest algorithmic breakthroughs in the field using a unified notation as well as a review of existing benchmarks' properties and shortcomings. Additionally, we provide a figure that summarizes the performance of each method and class of methods on different dataset properties, equipping researchers with the tools to decide which type of algorithm is best suited for the problem at hand and identify which classes of algorithms look the most promising. Finally, we provide our perspective on open problems and propose future research directions for this rapidly growing field.

Traveling Salesman Problem (TSP), as a classic routing optimization problem originally arising in the domain of transportation and logistics, has become a critical task in broader domains, such as manufacturing and biology. Recently, Deep Reinforcement Learning (DRL) has been increasingly employed to solve TSP due to its high inference efficiency. Nevertheless, most of existing end-to-end DRL algorithms only perform well on small TSP instances and can hardly generalize to large scale because of the drastically soaring memory consumption and computation time along with the enlarging problem scale. In this paper, we propose a novel end-to-end DRL approach, referred to as Pointerformer, based on multi-pointer Transformer. Particularly, Pointerformer adopts both reversible residual network in the encoder and multi-pointer network in the decoder to effectively contain memory consumption of the encoder-decoder architecture. To further improve the performance of TSP solutions, Pointerformer employs both a feature augmentation method to explore the symmetries of TSP at both training and inference stages as well as an enhanced context embedding approach to include more comprehensive context information in the query. Extensive experiments on a randomly generated benchmark and a public benchmark have shown that, while achieving comparative results on most small-scale TSP instances as SOTA DRL approaches do, Pointerformer can also well generalize to large-scale TSPs.