In human-centric tasks such as healthcare and education, the heterogeneity among patients and students necessitates personalized treatments and instructional interventions. While reinforcement learning (RL) has been utilized in those tasks, off-policy selection (OPS) is pivotal to close the loop by offline evaluating and selecting policies without online interactions, yet current OPS methods often overlook the heterogeneity among participants. Our work is centered on resolving a pivotal challenge in human-centric systems (HCSs): how to select a policy to deploy when a new participant joining the cohort, without having access to any prior offline data collected over the participant? We introduce First-Glance Off-Policy Selection (FPS), a novel approach that systematically addresses participant heterogeneity through sub-group segmentation and tailored OPS criteria to each sub-group. By grouping individuals with similar traits, FPS facilitates personalized policy selection aligned with unique characteristics of each participant or group of participants. FPS is evaluated via two important but challenging applications, intelligent tutoring systems and a healthcare application for sepsis treatment and intervention. FPS presents significant advancement in enhancing learning outcomes of students and in-hospital care outcomes.

We propose an extension of the reinforcement learning architecture that enables moral decision-making of reinforcement learning agents based on normative reasons. Central to this approach is a reason-based shield generator yielding a moral shield that binds the agent to actions that conform with recognized normative reasons so that our overall architecture restricts the agent to actions that are (internally) morally justified. In addition, we describe an algorithm that allows to iteratively improve the reason-based shield generator through case-based feedback from a moral judge.

Soft robotic manipulators offer operational advantage due to their compliant and deformable structures. However, their inherently nonlinear dynamics presents substantial challenges. Traditional analytical methods often depend on simplifying assumptions, while learning-based techniques can be computationally demanding and limit the control policies to existing data. This paper introduces a novel approach to soft robotic control, leveraging state-of-the-art policy gradient methods within parallelizable synthetic environments learned from data. We also propose a safety oriented actuation space exploration protocol via cascaded updates and weighted randomness. Specifically, our recurrent forward dynamics model is learned by generating a training dataset from a physically safe \textit{mean reverting} random walk in actuation space to explore the partially-observed state-space. We demonstrate a reinforcement learning approach towards closed-loop control through state-of-the-art actor-critic methods, which efficiently learn high-performance behaviour over long horizons. This approach removes the need for any knowledge regarding the robot's operation or capabilities and sets the stage for a comprehensive benchmarking tool in soft robotics control.

The autonomous exploration of environments by multi-robot systems is a critical task with broad applications in rescue missions, exploration endeavors, and beyond. Current approaches often rely on either greedy frontier selection or end-to-end deep reinforcement learning (DRL) methods, yet these methods are frequently hampered by limitations such as short-sightedness, overlooking long-term implications, and convergence difficulties stemming from the intricate high-dimensional learning space. To address these challenges, this paper introduces an innovative integration strategy that combines the low-dimensional action space efficiency of frontier-based methods with the far-sightedness and optimality of DRL-based approaches. We propose a three-tiered planning framework that first identifies frontiers in free space, creating a sparse map representation that lightens data transmission burdens and reduces the DRL action space's dimensionality. Subsequently, we develop a multi-graph neural network (mGNN) that incorporates states of potential targets and robots, leveraging policy-based reinforcement learning to compute affinities, thereby superseding traditional heuristic utility values. Lastly, we implement local routing planning through subsequence search, which avoids exhaustive sequence traversal. Extensive validation across diverse scenarios and comprehensive simulation results demonstrate the effectiveness of our proposed method. Compared to baseline approaches, our framework achieves environmental exploration with fewer time steps and a notable reduction of over 30% in data transmission, showcasing its superiority in terms of efficiency and performance.

The emergence of 3D Gaussian Splatting for fast and high-quality novel view synthesize has opened up the possibility to construct photo-realistic simulations from video for robotic reinforcement learning. While the approach has been demonstrated in several research papers, the software tools used to build such a simulator remain unavailable or proprietary. We present SplatGym, an open source neural simulator for training data-driven robotic control policies. The simulator creates a photorealistic virtual environment from a single video. It supports ego camera view generation, collision detection, and virtual object in-painting. We demonstrate training several visual navigation policies via reinforcement learning. SplatGym represents a notable first step towards an open-source general-purpose neural environment for robotic learning. It broadens the range of applications that can effectively utilise reinforcement learning by providing convenient and unrestricted tooling, and by eliminating the need for the manual development of conventional 3D environments.

In this work, we study the problem of finding Pareto optimal policies in multi-agent reinforcement learning problems with cooperative reward structures. We show that any algorithm where each agent only optimizes their reward is subject to suboptimal convergence. Therefore, to achieve Pareto optimality, agents have to act altruistically by considering the rewards of others. This observation bridges the multi-objective optimization framework and multi-agent reinforcement learning together. We first propose a framework for applying the Multiple Gradient Descent algorithm (MGDA) for learning in multi-agent settings. We further show that standard MGDA is subjected to weak Pareto convergence, a problem that is often overlooked in other learning settings but is prevalent in multi-agent reinforcement learning. To mitigate this issue, we propose MGDA++, an improvement of the existing algorithm to handle the weakly optimal convergence of MGDA properly. Theoretically, we prove that MGDA++ converges to strong Pareto optimal solutions in convex, smooth bi-objective problems. We further demonstrate the superiority of our MGDA++ in cooperative settings in the Gridworld benchmark. The results highlight that our proposed method can converge efficiently and outperform the other methods in terms of the optimality of the convergent policies. The source code is available at \url{https://github.com/giangbang/Strong-Pareto-MARL}.

Shared control problems involve a robot learning to collaborate with a human. When learning a shared control policy, short communication between the agents can often significantly reduce running times and improve the system's accuracy. We extend the shared control problem to include the ability to directly query a cooperating agent. We consider two types of potential responses to a query, namely oracles: one that can provide the learner with the best action they should take, even when that action might be myopically wrong, and one with a bounded knowledge limited to its part of the system. Given this additional information channel, this work further presents three heuristics for choosing when to query: reinforcement learning-based, utility-based, and entropy-based. These heuristics aim to reduce a system's overall learning cost. Empirical results on two environments show the benefits of querying to learn a better control policy and the tradeoffs between the proposed heuristics.

We study infinite-horizon average-reward reinforcement learning (RL) for Lipschitz MDPs and develop an algorithm ZoRL that discretizes the state-action space adaptively and zooms into promising regions of the state-action space. We show that its regret can be bounded as $\mathcal{\tilde{O}}\big(T^{1 - d_{\text{eff.}}^{-1}}\big)$, where $d_{\text{eff.}} = 2d_\mathcal{S} + d_z + 3$, $d_\mathcal{S}$ is the dimension of the state space, and $d_z$ is the zooming dimension. $d_z$ is a problem-dependent quantity, which allows us to conclude that if MDP is benign, then its regret will be small. We note that the existing notion of zooming dimension for average reward RL is defined in terms of policy coverings, and hence it can be huge when the policy class is rich even though the underlying MDP is simple, so that the regret upper bound is nearly $O(T)$. The zooming dimension proposed in the current work is bounded above by $d$, the dimension of the state-action space, and hence is truly adaptive, i.e., shows how to capture adaptivity gains for infinite-horizon average-reward RL. ZoRL outperforms other state-of-the-art algorithms in experiments; thereby demonstrating the gains arising due to adaptivity.

Image and video generative models that are pre-trained on Internet-scale data can greatly increase the generalization capacity of robot learning systems. These models can function as high-level planners, generating intermediate subgoals for low-level goal-conditioned policies to reach. However, the performance of these systems can be greatly bottlenecked by the interface between generative models and low-level controllers. For example, generative models may predict photorealistic yet physically infeasible frames that confuse low-level policies. Low-level policies may also be sensitive to subtle visual artifacts in generated goal images. This paper addresses these two facets of generalization, providing an interface to effectively "glue together" language-conditioned image or video prediction models with low-level goal-conditioned policies. Our method, Generative Hierarchical Imitation Learning-Glue (GHIL-Glue), filters out subgoals that do not lead to task progress and improves the robustness of goal-conditioned policies to generated subgoals with harmful visual artifacts. We find in extensive experiments in both simulated and real environments that GHIL-Glue achieves a 25% improvement across several hierarchical models that leverage generative subgoals, achieving a new state-of-the-art on the CALVIN simulation benchmark for policies using observations from a single RGB camera. GHIL-Glue also outperforms other generalist robot policies across 3/4 language-conditioned manipulation tasks testing zero-shot generalization in physical experiments.

Dual-arm manipulation is an area of growing interest in the robotics community. Enabling robots to perform tasks that require the coordinated use of two arms, is essential for complex manipulation tasks such as handling large objects, assembling components, and performing human-like interactions. However, achieving effective dual-arm manipulation is challenging due to the need for precise coordination, dynamic adaptability, and the ability to manage interaction forces between the arms and the objects being manipulated. We propose a novel pipeline that combines the advantages of policy learning based on environment feedback and gradient-based optimization to learn controller gains required for the control outputs. This allows the robotic system to dynamically modulate its impedance in response to task demands, ensuring stability and dexterity in dual-arm operations. We evaluate our pipeline on a trajectory-tracking task involving a variety of large, complex objects with different masses and geometries. The performance is then compared to three other established methods for controlling dual-arm robots, demonstrating superior results.