Robot learning from humans has been proposed and researched for several decades as a means to enable robots to learn new skills or adapt existing ones to new situations. Recent advances in AI, including learning approaches like reinforcement learning and architectures like transformers and foundation models, combined with access to massive datasets, have created attractive opportunities to apply those data-hungry techniques to this problem. We argue that the focus on massive amounts of pre-collected data, and the resulting learning paradigm, where humans demonstrate and robots learn in isolation, is overshadowing a specialized area of work we term Human-Interactive Robot Learning (HIRL). This paradigm, wherein robots and humans interact during the learning process, is at the intersection of multiple fields (AI, robotics, human-computer interaction, design and others) and holds unique promise. Using HIRL, robots can achieve greater sample efficiency (as humans can provide task knowledge through interaction), align with human preferences (as humans can guide the robot behavior toward their expectations), and explore more meaningfully and safely (as humans can utilize domain knowledge to guide learning and prevent catastrophic failures). This can result in robotic systems that can more quickly and easily adapt to new tasks in human environments. The objective of this article is to provide a broad and consistent overview of HIRL research and to guide researchers toward understanding the scope of HIRL, and current open or underexplored challenges related to four themes-namely, human, robot learning, interaction, and broader context. The article includes concrete use cases to illustrate the interaction between these challenges and inspire further research according to broad recommendations and a call for action for the growing HIRL community.

Reward design remains a significant bottleneck in applying reinforcement learning (RL) to real-world problems. A popular alternative is reward learning, where reward functions are inferred from human feedback rather than manually specified. Recent work has proposed learning reward functions from human ratings rather than traditional binary preferences, enabling richer and potentially less cognitively demanding supervision. Building on this paradigm, we introduce a new rating-based RL method, Ranked Return Regression for RL (R4). At its core, R4 uses a novel ranking mean squared error loss that learns from a dataset of trajectory–rating pairs, treating the human-provided discrete ratings (e.g., "bad," "neutral," "good") as ordinal targets. Unlike prior rating-based approaches, R4 offers formal guarantees: its solution set is provably minimal and complete under mild assumptions. Empirically, using both human-provided and simulated ratings, we demonstrate that R4 consistently matches or outperforms existing rating and preference-based RL methods on robotic benchmarks from OpenAI Gym and the DeepMind Control Suite.

3D-IC netlist partitioning is commonly optimized using proxy objectives, while final PPA is treated as a costly evaluation rather than an optimization signal. This proxy-driven paradigm makes it difficult to reliably translate additional PPA evaluations into better PPA outcomes. To bridge this gap, we present DOPP (D-Optimal PPA-driven partitioning selection), an approach that bridges the gap between proxies and true PPA metrics. Across eight 3D-IC designs, our framework improves PPA over Open3DBench (average relative improvements of 9.99

Scientific peer review faces mounting strain as submission volumes surge, making it increasingly difficult to sustain review quality, consistency, and timeliness. Recent advances in AI have led the community to consider its use in peer review, yet a key unresolved question is whether AI can generate technically sound reviews at real-world conference scale. Here we report the first large-scale field deployment of AI-assisted peer review: every main-track submission at AAAI-26 received one clearly identified AI review from a state-of-the-art system. The system combined frontier models, tool use, and safeguards in a multi-stage process to generate reviews for all 22,977 full-review papers in less than a day. A large-scale survey of AAAI-26 authors and program committee members showed that participants not only found AI reviews useful, but actually preferred them to human reviews on key dimensions such as technical accuracy and research suggestions. We also introduce a novel benchmark and find that our system substantially outperforms a simple LLM-generated review baseline at detecting a variety of scientific weaknesses. Together, these results show that state-of-the-art AI methods can already make meaningful contributions to scientific peer review at conference scale, opening a path toward the next generation of synergistic human-AI teaming for evaluating research.

To create useful reinforcement learning (RL) agents, step zero is to design a suitable reward function that captures the nuances of the task. However, reward engineering can be a difficult and time-consuming process. Instead, human-in-the-loop (HitL) RL approaches allow agents to learn reward functions from human feedback. Despite recent successes, many of the HitL RL methods still require numerous human interactions to learn successful reward functions. To that end, this work introduces Sub-optimal Data Pre-training, SDP, a method that leverages reward-free, sub-optimal data to improve the feedback efficiency of HitL RL algorithms. We demonstrate that SDP can significantly improve over state-of-the-art HitL RL algorithms in three DMControl environments.

ACM Reference Format: Chaitanya Kharyal, Tanmay Sinha, Sai KrishnaGottipati, FatemehAbdollahi, Srijita Das, and Matthew E. Taylor. 2023. Do As You Teach: A Multi-Teacher Approach to Self-Play in Deep Reinforcement Learning: Extended Abstract. In Proc. of the 22nd International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2023), London, United Kingdom, May 29 – June 2, 2023, IFAAMAS, 3 pages.

Deep Reinforcement Learning (RL) has achieved notable success across various domains. However, it still faces the sample-inefficiency problem that requires massive training samples to learn the optimal policy. Furthermore, the trained policy highly depends on the training environment, limiting the generalization, especially in long-horizon tasks. In this paper, we propose Plan-guided Exploration and generAlization for Reinforcement Learning (PEARL) to explore how symbolic planning can help Deep RL in terms of efficiency and generalization. PEARL is a two-level structure that incorporates any symbolic planner as the meta-controller to guide the low-level RL agent in finishing the long-horizon tasks. Furthermore, we automatically learn the mapping between the two levels based on a small dataset to relax the assumption of human efforts. We evaluate PEARL on Montezuma's Revenge and show that PEARL outperforms previous hierarchical and symbolic-planning-based methods in terms of efficiency and generalization.

Centralised training with decentralised execution (CT-DE) serves as the foundation of many leading multi-agent reinforcement learning (MARL) algorithms. Despite its popularity, it suffers from a critical drawback due to its reliance on learning from a single sample of the joint-action at a given state. As agents explore and update their policies during training, these single samples may poorly represent the actual joint-policy of the system of agents leading to high variance gradient estimates that hinder learning. To address this problem, we propose an enhancement tool that accommodates any actor-critic MARL method. Our framework, Performance Enhancing Reinforcement Learning Apparatus (PERLA), introduces a sampling technique of the agents' joint-policy into the critics while the agents train. This leads to TD updates that closely approximate the true expected value under the current joint-policy rather than estimates from a single sample of the joint-action at a given state. This produces low variance and precise estimates of expected returns, minimising the variance in the critic estimators which typically hinders learning. Moreover, as we demonstrate, by eliminating much of the critic variance from the single sampling of the joint policy, PERLA enables CT-DE methods to scale more efficiently with the number of agents. Theoretically, we prove that PERLA reduces variance in value estimates similar to that of decentralised training while maintaining the benefits of centralised training. Empirically, we demonstrate PERLA's superior performance and ability to reduce estimator variance in a range of benchmarks including Multi-agent Mujoco, and StarCraft II Multi-agent Challenge.

Morphology-aware policy learning is a means of enhancing policy sample efficiency by aggregating data from multiple agents. These types of policies have previously been shown to help generalize over dynamic, kinematic, and limb configuration variations between agent morphologies. Unfortunately, these policies still have sub-optimal zero-shot performance compared to end-to-end finetuning on morphologies at deployment. This limitation has ramifications in practical applications such as robotics because further data collection to perform end-to-end finetuning can be computationally expensive. In this work, we investigate combining morphology-aware pretraining with parameter efficient finetuning (PEFT) techniques to help reduce the learnable parameters necessary to specialize a morphology-aware policy to a target embodiment. We compare directly tuning sub-sets of model weights, input learnable adapters, and prefix tuning techniques for online finetuning. Our analysis reveals that PEFT techniques in conjunction with policy pre-training generally help reduce the number of samples to necessary to improve a policy compared to training models end-to-end from scratch. We further find that tuning as few as less than 1

Reinforcement learning (RL) has shown impressive results in sequential decision-making tasks. Large Language Models (LLMs) and Vision-Language Models (VLMs) have recently emerged, exhibiting impressive capabilities in multimodal understanding and reasoning. These advances have led to a surge of research integrating LLMs and VLMs into RL. This survey reviews representative works in which LLMs and VLMs are used to overcome key challenges in RL, such as lack of prior knowledge, long-horizon planning, and reward design. We present a taxonomy that categorizes these LLM/VLM-assisted RL approaches into three roles: agent, planner, and reward. We conclude by exploring open problems, including grounding, bias mitigation, improved representations, and action advice. By consolidating existing research and identifying future directions, this survey establishes a framework for integrating LLMs and VLMs into RL, advancing approaches that unify natural language and visual understanding with sequential decision-making.

To integrate into human-centered environments, autonomous agents must learn from and adapt to humans in their native settings. Preference-based reinforcement learning (PbRL) can enable this by learning reward functions from human preferences. However, humans live in a world full of diverse information, most of which is irrelevant to completing any particular task. It then becomes essential that agents learn to focus on the subset of task-relevant state features. To that end, this work proposes R2N (Robust-to-Noise), the first PbRL algorithm that leverages principles of dynamic sparse training to learn robust reward models that can focus on task-relevant features. In experiments with a simulated teacher, we demonstrate that R2N can adapt the sparse connectivity of its neural networks to focus on task-relevant features, enabling R2N to significantly outperform several sparse training and PbRL algorithms across simulated robotic environments.

Recent advances in reinforcement learning (RL) and Human-in-the-Loop (HitL) learning have made human-AI collaboration easier for humans to team with AI agents. Leveraging human expertise and experience with AI in intelligent systems can be efficient and beneficial. Still, it is unclear to what extent human-AI collaboration will be successful, and how such teaming performs compared to humans or AI agents only. In this work, we show that learning from humans is effective and that human-AI collaboration outperforms human-controlled and fully autonomous AI agents in a complex simulation environment. In addition, we have developed a new simulator for critical infrastructure protection, focusing on a scenario where AI-powered drones and human teams collaborate to defend an airport against enemy drone attacks. We develop a user interface to allow humans to assist AI agents effectively. We demonstrated that agents learn faster while learning from policy correction compared to learning from humans or agents. Furthermore, human-AI collaboration requires lower mental and temporal demands, reduces human effort, and yields higher performance than if humans directly controlled all agents. In conclusion, we show that humans can provide helpful advice to the RL agents, allowing them to improve learning in a multi-agent setting.

With the growing popularity of deep reinforcement learning (DRL), human-in-the-loop (HITL) approach has the potential to revolutionize the way we approach decision-making problems and create new opportunities for human-AI collaboration. In this article, we introduce a novel multi-layered hierarchical HITL DRL algorithm that comprises three types of learning: self learning, imitation learning and transfer learning. In addition, we consider three forms of human inputs: reward, action and demonstration. Furthermore, we discuss main challenges, trade-offs and advantages of HITL in solving complex problems and how human information can be integrated in the AI solution systematically. To verify our technical results, we present a real-world unmanned aerial vehicles (UAV) problem wherein a number of enemy drones attack a restricted area. The objective is to design a scalable HITL DRL algorithm for ally drones to neutralize the enemy drones before they reach the area. To this end, we first implement our solution using an award-winning open-source HITL software called Cogment. We then demonstrate several interesting results such as (a) HITL leads to faster training and higher performance, (b) advice acts as a guiding direction for gradient methods and lowers variance, and (c) the amount of advice should neither be too large nor too small to avoid over-training and under-training. Finally, we illustrate the role of human-AI cooperation in solving two real-world complex scenarios, i.e., overloaded and decoy attacks.

Reinforcement learning agents are fundamentally limited by the quality of the reward functions they learn from, yet reward design is often overlooked under the assumption that a well-defined reward is readily available. However, in practice, designing rewards is difficult, and even when specified, evaluating their correctness is equally problematic: how do we know if a reward function is correctly specified? In our work, we address these challenges by focusing on reward alignment – assessing whether a reward function accurately encodes the preferences of a human stakeholder. As a concrete measure of reward alignment, we introduce the Trajectory Alignment Coefficient to quantify the similarity between a human stakeholder's ranking of trajectory distributions and those induced by a given reward function. We show that the Trajectory Alignment Coefficient exhibits desirable properties, such as not requiring access to a ground truth reward, invariance to potential-based reward shaping, and applicability to online RL. Additionally, in an 11 – person user study of RL practitioners, we found that access to the Trajectory Alignment Coefficient during reward selection led to statistically significant improvements. Compared to relying only on reward functions, our metric reduced cognitive workload by 1.5x, was preferred by 82

Social skills training targets behaviors necessary for success in social interactions. However, traditional classroom training for such skills is often insufficient to teach effective communication - one-to-one interaction in real-world scenarios is preferred to lecture-style information delivery. This paper introduces a framework that allows instructors to collaborate with large language models to dynamically design realistic scenarios for students to communicate. Our framework uses these scenarios to enable student rehearsal, provide immediate feedback, and visualize performance for both students and instructors. Unlike traditional intelligent tutoring systems, instructors can easily co-create scenarios with a large language model without technical skills. Additionally, the system generates new scenario branches in real time when existing options do not fit the student's response.

Artificial Intelligence (AI) systems increasingly shape many aspects of daily life, influencing our jobs, finances, healthcare, and online content. This expansion has led to the rise of human–AI systems, where humans communicate, collaborate, or otherwise interact with AI, such as using AI outputs to make decisions. While these systems have shown potential to enhance human capabilities and improve performance on benchmarks, evidence suggests that they often underperform compared to AI‐only or human‐only approaches in experiments and real‐world applications. Here, we argue that human–AI systems should be developed with a greater emphasis on human‐centered factors—such as usability, fairness, trust, and user autonomy—within the algorithmic design and evaluation process. We advocate for integrating human‐centered principles into AI development through human‐centered algorithmic design and contextual evaluation with real users. Drawing on interdisciplinary research and our tutorial at two major AI conferences, we highlight examples and strategies for AI researchers and practitioners to embed these principles effectively. This work offers a systematic synthesis that integrates technical, practical, and ethical insights into a unified framework. Additionally, we highlight critical ethical considerations, including fairness, labor, privacy, and human agency to ensure that systems meet performance goals while serving broader societal interests. Through this work, we aim to inspire the field to embrace a truly human‐centered approach to algorithmic design and deployment.

The transition of control from autonomous systems to human drivers is critical in automated driving systems, particularly due to the out-of-the-loop (OOTL) circumstances that reduce driver readiness and increase reaction times. Existing takeover strategies are based on fixed time-based transitions, which fail to account for real-time driver performance variations. This paper proposes an adaptive transition strategy that dynamically adjusts the control authority based on both the time and tracking ability of the driver trajectory. Shared control is modeled as a cooperative differential game, where control authority is modulated through time-varying objective functions instead of blending control torques directly. To ensure a more natural takeover, a driver-specific state-tracking matrix is introduced, allowing the transition to align with individual control preferences. Multiple transition strategies are evaluated using a cumulative trajectory error metric. Human-in-the-loop control scenarios of the standardized ISO lane change maneuvers demonstrate that adaptive transitions reduce trajectory deviations and driver control effort compared to conventional strategies. Experiments also confirm that continuously adjusting control authority based on real-time deviations enhances vehicle stability while reducing driver effort during takeover.

Limited training capacity has contributed to a critical shortage of licensed commercial pilots. Adaptive educational technologies and simulators could alleviate current training bottlenecks if these technologies could assess trainee performance and provide appropriate feedback. Agents can be used to assess trainee performance, but there is insufficient guidance on how to provide concurrent feedback in simulation-based learning environments. So, we designed 4 feedback conditions that provide varying degrees of elaboration and used a within-subject study (n= 20) to compare feedback approaches. Trainee performance was best when they received highly-elaborative feedback that modeled expert behaviour. Variability in participant performance and preferences indicates a need to adapt the feedback type to individual learners and provides insight into the use of concurrent feedback in simulation-based learning environments. Specifically, learners appreciated the expert model because it facilitated a sense of control which was associated with lower negative affect and lower extraneous cognitive load.

A tenet of reinforcement learning is that the agent always observes rewards. However, this is not true in many realistic settings, e.g., a human observer may not always be available to provide rewards, sensors may be limited or malfunctioning, or rewards may be inaccessible during deployment. Monitored Markov decision processes (Mon-MDPs) have recently been proposed to model such settings. However, existing Mon-MDP algorithms have several limitations: they do not fully exploit the problem structure, cannot leverage a known monitor, lack worst-case guarantees for 'unsolvable' Mon-MDPs without specific initialization, and offer only asymptotic convergence proofs. This paper makes three contributions. First, we introduce a model-based algorithm for Mon-MDPs that addresses these shortcomings. The algorithm employs two instances of model-based interval estimation: one to ensure that observable rewards are reliably captured, and another to learn the minimax-optimal policy. Second, we empirically demonstrate the advantages. We show faster convergence than prior algorithms in over four dozen benchmarks, and even more dramatic improvement when the monitoring process is known. Third, we present the first finite-sample bound on performance. We show convergence to a minimax-optimal policy even when some rewards are never observable.