Reward machines have shown great promise at capturing non-Markovian reward functions for learning tasks that involve complex action sequencing. However, no algorithm currently exists for learning reward machines with realistic weak feedback in the form of preferences. We contribute REMAP, a novel algorithm for learning reward machines from preferences, with correctness and termination guarantees. REMAP introduces preference queries in place of membership queries in the L* algorithm, and leverages a symbolic observation table along with unification and constraint solving to narrow the hypothesis reward machine search space. In addition to the proofs of correctness and termination for REMAP, we present empirical evidence measuring correctness: how frequently the resulting reward machine is isomorphic under a consistent yet inexact teacher, and the regret between the ground truth and learned reward machines.

We propose Reinforcement Learning from Contrast Distillation (RLCD), a method for aligning language models to follow natural language principles without using human feedback. RLCD trains a preference model using simulated preference pairs that contain both a high-quality and low-quality example, generated using contrasting positive and negative prompts. The preference model is then used to improve a base unaligned language model via reinforcement learning. Empirically, RLCD outperforms RLAIF (Bai et al., 2022b) and context distillation (Huang et al., 2022) baselines across three diverse alignment tasks--harmlessness, helpfulness, and story outline generation--and on both 7B and 30B model scales for preference data simulation. Reinforcement Learning from Human Feedback (RLHF) has recently been used to great effect to align pretrained large language models (LLMs) to human preferences, optimizing for desirable qualities like harmlessness and helpfulness (Bai et al., 2022a) and achieving ...

Reinforcement learning is the classic paradigm for addressing sequential decision-making problems [47]. Naturally, while inheriting the fundamental challenge of generalization across novel states and actions from supervised learning, general-purpose reinforcement-learning agents must also contend with the additional challenges of exploration and credit assignment. While much initial progress in the field was driven largely by innovative machinery for tackling credit assignment [45, 46, 44] alongside simple exploration heuristics (ε-greedy exploration, for example), recent years have seen a reversal with the bulk of attention focused on a broad array of exploration methods (spanning additional heuristics as well as more principled approaches) [51, 38, 15], and relatively little consideration given to issues of credit assignment. This lack of interest in solution concepts, however, has not stopped the proliferation of reinforcement learning into novel application areas characterized by long problem horizons and sparse reward signals; indeed, the current reinforcement learning from human feedback (RLHF) paradigm [28] is now a widely popularized example of an environment that operates in perhaps the harshest setting where a single feedback signal is only obtained after the completion of a long trajectory.

Developing and testing automated driving models in the real world might be challenging and even dangerous, while simulation can help with this, especially for challenging maneuvers. Deep reinforcement learning (DRL) has the potential to tackle complex decision-making and controlling tasks through learning and interacting with the environment, thus it is suitable for developing automated driving while not being explored in detail yet. This study carried out a comprehensive study by implementing, evaluating, and comparing the two DRL algorithms, Deep Q-networks (DQN) and Trust Region Policy Optimization (TRPO), for training automated driving on the highway-env simulation platform. Effective and customized reward functions were developed and the implemented algorithms were evaluated in terms of onlane accuracy (how well the car drives on the road within the lane), efficiency (how fast the car drives), safety (how likely the car is to crash into obstacles), and comfort (how much the car makes jerks, e.g., suddenly accelerates or brakes). Results show that the TRPO-based models with modified reward functions delivered the best performance in most cases. Furthermore, to train a uniform driving model that can tackle various driving maneuvers besides the specific ones, this study expanded the highway-env and developed an extra customized training environment, namely, ComplexRoads, integrating various driving maneuvers and multiple road scenarios together. Models trained on the designed ComplexRoads environment can adapt well to other driving maneuvers with promising overall performance. Lastly, several functionalities were added to the highway-env to implement this work. The codes are open on GitHub at https://github.com/alaineman/drlcarsim-paper.

Current frameworks for training offensive penetration testing agents with deep reinforcement learning struggle to produce agents that perform well in real-world scenarios, due to the reality gap in simulation-based frameworks and the lack of scalability in emulation-based frameworks. Additionally, existing frameworks often use an unrealistic metric that measures the agents' performance on the training data. NASimEmu, a new framework introduced in this paper, addresses these issues by providing both a simulator and an emulator with a shared interface. This approach allows agents to be trained in simulation and deployed in the emulator, thus verifying the realism of the used abstraction. Our framework promotes the development of general agents that can transfer to novel scenarios unseen during their training. For the simulation part, we adopt an existing simulator NASim and enhance its realism. The emulator is implemented with industry-level tools, such as Vagrant, VirtualBox, and Metasploit. Experiments demonstrate that a simulation-trained agent can be deployed in emulation, and we show how to use the framework to train a general agent that transfers into novel, structurally different scenarios. NASimEmu is available as open-source.

Multi-agent reinforcement learning (MARL) is a powerful tool for training automated systems acting independently in a common environment. However, it can lead to sub-optimal behavior when individual incentives and group incentives diverge. Humans are remarkably capable at solving these social dilemmas. It is an open problem in MARL to replicate such cooperative behaviors in selfish agents. In this work, we draw upon the idea of formal contracting from economics to overcome diverging incentives between agents in MARL. We propose an augmentation to a Markov game where agents voluntarily agree to binding state-dependent transfers of reward, under pre-specified conditions. Our contributions are theoretical and empirical. First, we show that this augmentation makes all subgame-perfect equilibria of all fully observed Markov games exhibit socially optimal behavior, given a sufficiently rich space of contracts. Next, we complement our game-theoretic analysis by showing that state-of-the-art RL algorithms learn socially optimal policies given our augmentation. Our experiments include classic static dilemmas like Stag Hunt, Prisoner's Dilemma and a public goods game, as well as dynamic interactions that simulate traffic, pollution management and common pool resource management.

Multi-objective Markov decision processes are a special kind of multi-objective optimization problem that involves sequential decision making while satisfying the Markov property of stochastic processes. Multi-objective reinforcement learning methods address this problem by fusing the reinforcement learning paradigm with multi-objective optimization techniques. One major drawback of these methods is the lack of adaptability to non-stationary dynamics in the environment. This is because they adopt optimization procedures that assume stationarity to evolve a coverage set of policies that can solve the problem. This paper introduces a developmental optimization approach that can evolve the policy coverage set while exploring the preference space over the defined objectives in an online manner. We propose a novel multi-objective reinforcement learning algorithm that can robustly evolve a convex coverage set of policies in an online manner in non-stationary environments. We compare the proposed algorithm with two state-of-the-art multi-objective reinforcement learning algorithms in stationary and non-stationary environments. Results showed that the proposed algorithm significantly outperforms the existing algorithms in non-stationary environments while achieving comparable results in stationary environments.

Multi-objective Markov decision processes are sequential decision-making problems that involve multiple conflicting reward functions that cannot be optimized simultaneously without a compromise. This type of problems cannot be solved by a single optimal policy as in the conventional case. Alternatively, multi-objective reinforcement learning methods evolve a coverage set of optimal policies that can satisfy all possible preferences in solving the problem. However, many of these methods cannot generalize their coverage sets to work in non-stationary environments. In these environments, the parameters of the state transition and reward distribution vary over time. This limitation results in significant performance degradation for the evolved policy sets. In order to overcome this limitation, there is a need to learn a generic skill set that can bootstrap the evolution of the policy coverage set for each shift in the environment dynamics therefore, it can facilitate a continuous learning process. In this work, intrinsically motivated reinforcement learning has been successfully deployed to evolve generic skill sets for learning hierarchical policies to solve multi-objective Markov decision processes. We propose a novel dual-phase intrinsically motivated reinforcement learning method to address this limitation. In the first phase, a generic set of skills is learned. While in the second phase, this set is used to bootstrap policy coverage sets for each shift in the environment dynamics. We show experimentally that the proposed method significantly outperforms state-of-the-art multi-objective reinforcement methods in a dynamic robotics environment.

Urban Air Mobility (UAM) promises a new dimension to decongested, safe, and fast travel in urban and suburban hubs. These UAM aircraft are conceived to operate from small airports called vertiports each comprising multiple take-off/landing and battery-recharging spots. Since they might be situated in dense urban areas and need to handle many aircraft landings and take-offs each hour, managing this schedule in real-time becomes challenging for a traditional air-traffic controller but instead calls for an automated solution. This paper provides a novel approach to this problem of Urban Air Mobility - Vertiport Schedule Management (UAM-VSM), which leverages graph reinforcement learning to generate decision-support policies. Here the designated physical spots within the vertiport's airspace and the vehicles being managed are represented as two separate graphs, with feature extraction performed through a graph convolutional network (GCN). Extracted features are passed onto perceptron layers to decide actions such as continue to hover or cruise, continue idling or take-off, or land on an allocated vertiport spot. Performance is measured based on delays, safety (no. of collisions) and battery consumption. Through realistic simulations in AirSim applied to scaled down multi-rotor vehicles, our results demonstrate the suitability of using graph reinforcement learning to solve the UAM-VSM problem and its superiority to basic reinforcement learning (with graph embeddings) or random choice baselines.

Synthesizing physically plausible human motions in 3D scenes is a challenging problem. Kinematics-based methods cannot avoid inherent artifacts (e.g., penetration and foot skating) due to the lack of physical constraints. Meanwhile, existing physics-based methods cannot generalize to multi-object scenarios since the policy trained with reinforcement learning has limited modeling capacity. In this work, we present a framework that enables physically simulated characters to perform long-term interaction tasks in diverse, cluttered, and unseen scenes. The key idea is to decompose human-scene interactions into two fundamental processes, Interacting and Navigating, which motivates us to construct two reusable Controller, i.e., InterCon and NavCon. Specifically, InterCon contains two complementary policies that enable characters to enter and leave the interacting state (e.g., sitting on a chair and getting up). To generate interaction with objects at different places, we further design NavCon, a trajectory following policy, to keep characters' locomotion in the free space of 3D scenes. Benefiting from the divide and conquer strategy, we can train the policies in simple environments and generalize to complex multi-object scenes. Experimental results demonstrate that our framework can synthesize physically plausible long-term human motions in complex 3D scenes. Code will be publicly released at https://github.com/liangpan99/InterScene.