We consider an offline reinforcement learning (RL) setting where the agent need to learn from a dataset collected by rolling out multiple behavior policies. There are two challenges for this setting: 1) The optimal trade-off between optimizing the RL signal and the behavior cloning (BC) signal changes on different states due to the variation of the action coverage induced by different behavior policies. Previous methods fail to handle this by only controlling the global trade-off. 2) For a given state, the action distribution generated by different behavior policies may have multiple modes. The BC regularizers in many previous methods are mean-seeking, resulting in policies that select out-of-distribution (OOD) actions in the middle of the modes. In this paper, we address both challenges by using adaptively weighted reverse Kullback-Leibler (KL) divergence as the BC regularizer based on the TD3 algorithm. Our method not only trades off the RL and BC signals with per-state weights (i.e., strong BC regularization on the states with narrow action coverage, and vice versa) but also avoids selecting OOD actions thanks to the mode-seeking property of reverse KL. Empirically, our algorithm can outperform existing offline RL algorithms in the MuJoCo locomotion tasks with the standard D4RL datasets as well as the mixed datasets that combine the standard datasets.

Despite recent advances in RL research, the ability to generalize to new tasks remains one of the major issues in both reinforcement learning (RL) and decision-making. RL agents perform remarkably in a single-task setting but frequently make mistakes when faced with unforeseen obstacles. Additionally, single-task RL agents can largely overfit the tasks they are trained on, rendering them unsuitable for real-world applications. This is where a general agent that can successfully handle various unprecedented tasks and unforeseen difficulties can be useful. The vast majority of general agents are trained using a variety of diverse tasks.

In this paper, we present a solution to a design problem of control strategies for multi-agent cooperative transport. Although existing learning-based methods assume that the number of agents is the same as that in the training environment, the number might differ in reality considering that the robots' batteries may completely discharge, or additional robots may be introduced to reduce the time required to complete a task. Therefore, it is crucial that the learned strategy be applicable to scenarios wherein the number of agents differs from that in the training environment. In this paper, we propose a novel multi-agent reinforcement learning framework of event-triggered communication and consensus-based control for distributed cooperative transport. The proposed policy model estimates the resultant force and torque in a consensus manner using the estimates of the resultant force and torque with the neighborhood agents. Moreover, it computes the control and communication inputs to determine when to communicate with the neighboring agents under local observations and estimates of the resultant force and torque. Therefore, the proposed framework can balance the control performance and communication savings in scenarios wherein the number of agents differs from that in the training environment. We confirm the effectiveness of our approach by using a maximum of eight and six robots in the simulations and experiments, respectively.

Modern power systems will have to face difficult challenges in the years to come: frequent blackouts in urban areas caused by high power demand peaks, grid instability exacerbated by intermittent renewable generation, and global climate change amplified by rising carbon emissions. While current practices are growingly inadequate, the path to widespread adoption of artificial intelligence (AI) methods is hindered by missing aspects of trustworthiness. The CityLearn Challenge is an exemplary opportunity for researchers from multiple disciplines to investigate the potential of AI to tackle these pressing issues in the energy domain, collectively modeled as a reinforcement learning (RL) task. Multiple real-world challenges faced by contemporary RL techniques are embodied in the problem formulation. In this paper, we present a novel method using the solution function of optimization as policies to compute actions for sequential decision-making, while notably adapting the parameters of the optimization model from online observations. Algorithmically, this is achieved by an evolutionary algorithm under a novel trajectory-based guidance scheme. Formally, the global convergence property is established. Our agent ranked first in the latest 2021 CityLearn Challenge, being able to achieve superior performance in almost all metrics while maintaining some key aspects of interpretability.

Safety is still one of the major research challenges in reinforcement learning (RL). In this paper, we address the problem of how to avoid safety violations of RL agents during exploration in probabilistic and partially unknown environments. Our approach combines automata learning for Markov Decision Processes (MDPs) and shield synthesis in an iterative approach. Initially, the MDP representing the environment is unknown. The agent starts exploring the environment and collects traces. From the collected traces, we passively learn MDPs that abstractly represent the safety-relevant aspects of the environment. Given a learned MDP and a safety specification, we construct a shield. For each state-action pair within a learned MDP, the shield computes exact probabilities on how likely it is that executing the action results in violating the specification from the current state within the next $k$ steps. After the shield is constructed, the shield is used during runtime and blocks any actions that induce a too large risk from the agent. The shielded agent continues to explore the environment and collects new data on the environment. Iteratively, we use the collected data to learn new MDPs with higher accuracy, resulting in turn in shields able to prevent more safety violations. We implemented our approach and present a detailed case study of a Q-learning agent exploring slippery Gridworlds. In our experiments, we show that as the agent explores more and more of the environment during training, the improved learned models lead to shields that are able to prevent many safety violations.

Besides the recent impressive results on reinforcement learning (RL), safety is still one of the major research challenges in RL. RL is a machine-learning approach to determine near-optimal policies in Markov decision processes (MDPs). In this paper, we consider the setting where the safety-relevant fragment of the MDP together with a temporal logic safety specification is given and many safety violations can be avoided by planning ahead a short time into the future. We propose an approach for online safety shielding of RL agents. During runtime, the shield analyses the safety of each available action. For any action, the shield computes the maximal probability to not violate the safety specification within the next $k$ steps when executing this action. Based on this probability and a given threshold, the shield decides whether to block an action from the agent. Existing offline shielding approaches compute exhaustively the safety of all state-action combinations ahead of time, resulting in huge computation times and large memory consumption. The intuition behind online shielding is to compute at runtime the set of all states that could be reached in the near future. For each of these states, the safety of all available actions is analysed and used for shielding as soon as one of the considered states is reached. Our approach is well suited for high-level planning problems where the time between decisions can be used for safety computations and it is sustainable for the agent to wait until these computations are finished. For our evaluation, we selected a 2-player version of the classical computer game SNAKE. The game represents a high-level planning problem that requires fast decisions and the multiplayer setting induces a large state space, which is computationally expensive to analyse exhaustively.

Both goal-agnostic and goal-oriented tasks have practical value for robotic grasping: goal-agnostic tasks target all objects in the workspace, while goal-oriented tasks aim at grasping pre-assigned goal objects. However, most current grasping methods are only better at coping with one task. In this work, we propose a bifunctional push-grasping synergistic strategy for goal-agnostic and goal-oriented grasping tasks. Our method integrates pushing along with grasping to pick up all objects or pre-assigned goal objects with high action efficiency depending on the task requirement. We introduce a bifunctional network, which takes in visual observations and outputs dense pixel-wise maps of Q values for pushing and grasping primitive actions, to increase the available samples in the action space. Then we propose a hierarchical reinforcement learning framework to coordinate the two tasks by considering the goal-agnostic task as a combination of multiple goal-oriented tasks. To reduce the training difficulty of the hierarchical framework, we design a two-stage training method to train the two types of tasks separately. We perform pre-training of the model in simulation, and then transfer the learned model to the real world without any additional real-world fine-tuning. Experimental results show that the proposed approach outperforms existing methods in task completion rate and grasp success rate with less motion number. Supplementary material is available at https: //github.com/DafaRen/Learning_Bifunctional_Push-grasping_Synergistic_Strategy_for_Goal-agnostic_and_Goal-oriented_Tasks

Social bots are referred to as the automated accounts on social networks that make attempts to behave like human. While Graph Neural Networks (GNNs) has been massively applied to the field of social bot detection, a huge amount of domain expertise and prior knowledge is heavily engaged in the state-of-the art approaches to design a dedicated neural network architecture for a specific classification task. Involving oversized nodes and network layers in the model design, however, usually causes the over-smoothing problem and the lack of embedding discrimination. In this paper, we propose RoSGAS, a novel Reinforced and Self-supervised GNN Architecture Search framework to adaptively pinpoint the most suitable multi-hop neighborhood and the number of layers in the GNN architecture. More specifically, we consider the social bot detection problem as a user-centric subgraph embedding and classification task. We exploit heterogeneous information network to present the user connectivity by leveraging account metadata, relationships, behavioral features and content features. RoSGAS uses a multi-agent deep reinforcement learning (RL) mechanism for navigating the search of optimal neighborhood and network layers to learn individually the subgraph embedding for each target user. A nearest neighbor mechanism is developed for accelerating the RL training process, and RoSGAS can learn more discriminative subgraph embedding with the aid of self-supervised learning. Experiments on 5 Twitter datasets show that RoSGAS outperforms the state-of-the-art approaches in terms of accuracy, training efficiency and stability, and has better generalization when handling unseen samples.

Zero-sum Markov Games (MGs) has been an efficient framework for multi-agent systems and robust control, wherein a minimax problem is constructed to solve the equilibrium policies. At present, this formulation is well studied under tabular settings wherein the maximum operator is primarily and exactly solved to calculate the worst-case value function. However, it is non-trivial to extend such methods to handle complex tasks, as finding the maximum over large-scale action spaces is usually cumbersome. In this paper, we propose the smoothing policy iteration (SPI) algorithm to solve the zero-sum MGs approximately, where the maximum operator is replaced by the weighted LogSumExp (WLSE) function to obtain the nearly optimal equilibrium policies. Specially, the adversarial policy is served as the weight function to enable an efficient sampling over action spaces.We also prove the convergence of SPI and analyze its approximation error in $\infty -$norm based on the contraction mapping theorem. Besides, we propose a model-based algorithm called Smooth adversarial Actor-critic (SaAC) by extending SPI with the function approximations. The target value related to WLSE function is evaluated by the sampled trajectories and then mean square error is constructed to optimize the value function, and the gradient-ascent-descent methods are adopted to optimize the protagonist and adversarial policies jointly. In addition, we incorporate the reparameterization technique in model-based gradient back-propagation to prevent the gradient vanishing due to sampling from the stochastic policies. We verify our algorithm in both tabular and function approximation settings. Results show that SPI can approximate the worst-case value function with a high accuracy and SaAC can stabilize the training process and improve the adversarial robustness in a large margin.

We present an approach for safe trajectory planning, where a strategic task related to autonomous racing is learned sample-efficient within a simulation environment. A high-level policy, represented as a neural network, outputs a reward specification that is used within the cost function of a parametric nonlinear model predictive controller (NMPC). By including constraints and vehicle kinematics in the NLP, we are able to guarantee safe and feasible trajectories related to the used model. Compared to classical reinforcement learning (RL), our approach restricts the exploration to safe trajectories, starts with a good prior performance and yields full trajectories that can be passed to a tracking lowest-level controller. We do not address the lowest-level controller in this work and assume perfect tracking of feasible trajectories. We show the superior performance of our algorithm on simulated racing tasks that include high-level decision making. The vehicle learns to efficiently overtake slower vehicles and to avoid getting overtaken by blocking faster vehicles.