Trevor McInroe

Ibrahim H. Ahmed, Cillian Brewitt, Ignacio Carlucho et al. · microsoft-research

The development of autonomous agents which can interact with other agents to accomplish a given task is a core area of research in artificial intelligence and machine learning. Towards this goal, the Autonomous Agents Research Group develops novel machine learning algorithms for autonomous systems control, with a specific focus on deep reinforcement learning and multi-agent reinforcement learning. Research problems include scalable learning of coordinated agent policies and inter-agent communication; reasoning about the behaviours, goals, and composition of other agents from limited observations; and sample-efficient learning based on intrinsic motivation, curriculum learning, causal inference, and representation learning. This article provides a broad overview of the ongoing research portfolio of the group and discusses open problems for future directions.

Trevor McInroe, Lukas Schäfer, Stefano V. Albrecht · microsoft-research

Learning control from pixels is difficult for reinforcement learning (RL) agents because representation learning and policy learning are intertwined. Previous approaches remedy this issue with auxiliary representation learning tasks, but they either do not consider the temporal aspect of the problem or only consider single-step transitions, which may cause learning inefficiencies if important environmental changes take many steps to manifest. We propose Hierarchical $k$-Step Latent (HKSL), an auxiliary task that learns multiple representations via a hierarchy of forward models that learn to communicate and an ensemble of $n$-step critics that all operate at varying magnitudes of step skipping. We evaluate HKSL in a suite of 30 robotic control tasks with and without distractors and a task of our creation. We find that HKSL either converges to higher or optimal episodic returns more quickly than several alternative representation learning approaches. Furthermore, we find that HKSL's representations capture task-relevant details accurately across timescales (even in the presence of distractors) and that communication channels between hierarchy levels organize information based on both sides of the communication process, both of which improve sample efficiency.

Trevor McInroe, Adam Jelley, Stefano V. Albrecht et al.

Offline pretraining with a static dataset followed by online fine-tuning (offline-to-online, or OtO) is a paradigm well matched to a real-world RL deployment process. In this scenario, we aim to find the best-performing policy within a limited budget of online interactions. Previous work in the OtO setting has focused on correcting for bias introduced by the policy-constraint mechanisms of offline RL algorithms. Such constraints keep the learned policy close to the behavior policy that collected the dataset, but we show this can unnecessarily limit policy performance if the behavior policy is far from optimal. Instead, we forgo constraints and frame OtO RL as an exploration problem that aims to maximize the benefit of online data-collection. We first study the major online RL exploration methods based on intrinsic rewards and UCB in the OtO setting, showing that intrinsic rewards add training instability through reward-function modification, and UCB methods are myopic and it is unclear which learned-component's ensemble to use for action selection. We then introduce an algorithm for planning to go out-of-distribution (PTGOOD) that avoids these issues. PTGOOD uses a non-myopic planning procedure that targets exploration in relatively high-reward regions of the state-action space unlikely to be visited by the behavior policy. By leveraging concepts from the Conditional Entropy Bottleneck, PTGOOD encourages data collected online to provide new information relevant to improving the final deployment policy without altering rewards. We show empirically in several continuous control tasks that PTGOOD significantly improves agent returns during online fine-tuning and avoids the suboptimal policy convergence that many of our baselines exhibit in several environments.

Mhairi Dunion, Trevor McInroe, Kevin Sebastian Luck et al.

Reinforcement Learning (RL) agents are often unable to generalise well to environment variations in the state space that were not observed during training. This issue is especially problematic for image-based RL, where a change in just one variable, such as the background colour, can change many pixels in the image. The changed pixels can lead to drastic changes in the agent's latent representation of the image, causing the learned policy to fail. To learn more robust representations, we introduce TEmporal Disentanglement (TED), a self-supervised auxiliary task that leads to disentangled image representations exploiting the sequential nature of RL observations. We find empirically that RL algorithms utilising TED as an auxiliary task adapt more quickly to changes in environment variables with continued training compared to state-of-the-art representation learning methods. Since TED enforces a disentangled structure of the representation, our experiments also show that policies trained with TED generalise better to unseen values of variables irrelevant to the task (e.g. background colour) as well as unseen values of variables that affect the optimal policy (e.g. goal positions).

Ben Sanati, Thomas L. Lee, Trevor McInroe et al.

A fundamental challenge in developing general learning algorithms is their tendency to forget past knowledge when adapting to new data. Addressing this problem requires a principled understanding of forgetting; yet, despite decades of study, no unified definition has emerged that provides insights into the underlying dynamics of learning. We propose an algorithm- and task-agnostic theory that characterises forgetting as a lack of self-consistency in a learner's predictive distribution over future experiences, manifesting as a loss of predictive information. Our theory naturally yields a general measure of an algorithm's propensity to forget. To validate the theory, we design a comprehensive set of experiments that span classification, regression, generative modelling, and reinforcement learning. We empirically demonstrate how forgetting is present across all learning settings and plays a significant role in determining learning efficiency. Together, these results establish a principled understanding of forgetting and lay the foundation for analysing and improving the information retention capabilities of general learning algorithms.

Elle Miller, Jayaram Reddy, Ayush Deshmukh et al.

Tactile-based reinforcement learning (RL) is currently hindered by fragmented research and a focus on over-saturated orientation tasks. We introduce v2 of the Robot Tactile Olympiad (\texttt{roto 2.0}), a GPU-parallelised benchmark designed to standardise tactile-based RL across four distinct robotic morphologies (16-DOF to 24-DOF). Unlike prior benchmarks, roto focuses on end-to-end "blind" manipulation, utilising only proprioception and tactile sensing without state information or distillation. We demonstrate a significant performance leap, with our blind agents achieving 13 Baoding ball rotations in 10 seconds, an order of magnitude faster than current state-of-the-art speeds. By open-sourcing our environments and robustly tuned baselines, we reduce the barrier to entry and enable researchers to prioritise fundamental algorithmic challenges over tedious RL tuning. Website: https://elle-miller.github.io/roto/

Samuel Garcin, Trevor McInroe, Pablo Samuel Castro et al.

Extracting relevant information from a stream of high-dimensional observations is a central challenge for deep reinforcement learning agents. Actor-critic algorithms add further complexity to this challenge, as it is often unclear whether the same information will be relevant to both the actor and the critic. To this end, we here explore the principles that underlie effective representations for the actor and for the critic in on-policy algorithms. We focus our study on understanding whether the actor and critic will benefit from separate, rather than shared, representations. Our primary finding is that when separated, the representations for the actor and critic systematically specialise in extracting different types of information from the environment -- the actor's representation tends to focus on action-relevant information, while the critic's representation specialises in encoding value and dynamics information. We conduct a rigourous empirical study to understand how different representation learning approaches affect the actor and critic's specialisations and their downstream performance, in terms of sample efficiency and generation capabilities. Finally, we discover that a separated critic plays an important role in exploration and data collection during training. Our code, trained models and data are accessible at https://github.com/francelico/deac-rep.

Trevor McInroe, Samuel Garcin

We present PixelBrax, a set of continuous control tasks with pixel observations. We combine the Brax physics engine with a pure JAX renderer, allowing reinforcement learning (RL) experiments to run end-to-end on the GPU. PixelBrax can render observations over thousands of parallel environments and can run two orders of magnitude faster than existing benchmarks that rely on CPU-based rendering. Additionally, PixelBrax supports fully reproducible experiments through its explicit handling of any stochasticity within the environments and supports color and video distractors for benchmarking generalization. We open-source PixelBrax alongside JAX implementations of several RL algorithms at github.com/trevormcinroe/pixelbrax.

Leonard Hinckeldey, Elliot Fosong, Elle Miller et al.

The development of reinforcement learning (RL) algorithms has been largely driven by ambitious challenge tasks and benchmarks. Games have dominated RL benchmarks because they present relevant challenges, are inexpensive to run and easy to understand. While games such as Go and Atari have led to many breakthroughs, they often do not directly translate to real-world embodied applications. In recognising the need to diversify RL benchmarks and addressing complexities that arise in embodied interaction scenarios, we introduce Assistax: an open-source benchmark designed to address challenges arising in assistive robotics tasks. Assistax uses JAX's hardware acceleration for significant speed-ups for learning in physics-based simulations. In terms of open-loop wall-clock time, Assistax runs up to $370\times$ faster when vectorising training runs compared to CPU-based alternatives. Assistax conceptualises the interaction between an assistive robot and an active human patient using multi-agent RL to train a population of diverse partner agents against which an embodied robotic agent's zero-shot coordination capabilities can be tested. Extensive evaluation and hyperparameter tuning for popular continuous control RL and MARL algorithms provide reliable baselines and establish Assistax as a practical benchmark for advancing RL research for assistive robotics. The code is available at: https://github.com/assistive-autonomy/assistax.

Adam Jelley, Trevor McInroe, Sam Devlin et al.

Supervised imitation-based approaches are often favored over off-policy reinforcement learning approaches for learning policies offline, since their straightforward optimization objective makes them computationally efficient and stable to train. However, their performance is fundamentally limited by the behavior policy that collected the dataset. Off-policy reinforcement learning provides a promising approach for improving on the behavior policy, but training is often computationally inefficient and unstable due to temporal-difference bootstrapping. In this paper, we propose a best-of-both approach by pre-training with supervised learning before improving performance with off-policy reinforcement learning. Specifically, we demonstrate improved efficiency by pre-training an actor with behavior cloning and a critic with a supervised Monte-Carlo value error. We find that we are able to substantially improve the training time of popular off-policy algorithms on standard benchmarks, and also achieve greater stability. Code is available at: https://github.com/AdamJelley/EfficientOfflineRL

Dongge Han, Trevor McInroe, Adam Jelley et al.

Large language models (LLMs) have shown significant potential for robotics applications, particularly task planning, by harnessing their language comprehension and text generation capabilities. However, in applications such as household robotics, a critical gap remains in the personalization of these models to individual user preferences. We introduce LLM-Personalize, a novel framework with an optimization pipeline designed to personalize LLM planners for household robotics. Our LLM-Personalize framework features an LLM planner that performs iterative planning in multi-room, partially-observable household scenarios, making use of a scene graph constructed with local observations. The generated plan consists of a sequence of high-level actions which are subsequently executed by a controller. Central to our approach is the optimization pipeline, which combines imitation learning and iterative self-training to personalize the LLM planner. In particular, the imitation learning phase performs initial LLM alignment from demonstrations, and bootstraps the model to facilitate effective iterative self-training, which further explores and aligns the model to user preferences. We evaluate LLM-Personalize on Housekeep, a challenging simulated real-world 3D benchmark for household rearrangements, and show that LLM-Personalize achieves more than a 30 percent increase in success rate over existing LLM planners, showcasing significantly improved alignment with human preferences. Project page: https://gdg94.github.io/projectllmpersonalize/.

Marcel Hedman, Kale-ab Abebe Tessera, Juan Claude Formanek et al.

Offline multi-agent reinforcement learning (MARL) enables policy learning from fixed datasets, but is prone to coordination failure: agents trained on static, off-policy data converge to suboptimal joint behaviours because they cannot co-adapt as their policies change. We introduce CODA (Coordination via On-Policy Diffusion for Multi-Agent Reinforcement Learning), a diffusion-based multi-agent trajectory generator for data augmentation that samples conditioned on the current joint policy, producing synthetic experience which reflects the evolving behaviours of the agents, thereby providing a mechanism for co-adaptation. We find that previous diffusion-based augmentation approaches are insufficient for fostering multi-agent coordination because they produce static augmented datasets that do not evolve as the current joint policy changes during training; CODA resolves this by more closely simulating on-policy learning and is a meaningful step toward coordinated behaviours in the offline setting. CODA is algorithm-agnostic and can be layered onto both model-free and model-based offline reinforcement learning pipelines as an augmentation module. Empirically, CODA not only resolves canonical coordination pathologies in continuous polynomial games but also delivers strong results on the more complex MaMuJoCo continuous-control benchmarks.

Weipu Zhang, Adam Jelley, Trevor McInroe et al.

Deep reinforcement learning has achieved remarkable success in learning control policies from pixels across a wide range of tasks, yet its application remains hindered by low sample efficiency, requiring significantly more environment interactions than humans to reach comparable performance. Model-based reinforcement learning (MBRL) offers a solution by leveraging learnt world models to generate simulated experience, thereby improving sample efficiency. However, in visually complex environments, small or dynamic elements can be critical for decision-making. Yet, traditional MBRL methods in pixel-based environments typically rely on auto-encoding with an $L_2$ loss, which is dominated by large areas and often fails to capture decision-relevant details. To address these limitations, we propose an object-centric MBRL pipeline, which integrates recent advances in computer vision to allow agents to focus on key decision-related elements. Our approach consists of four main steps: (1) annotating key objects related to rewards and goals with segmentation masks, (2) extracting object features using a pre-trained, frozen foundation vision model, (3) incorporating these object features with the raw observations to predict environmental dynamics, and (4) training the policy using imagined trajectories generated by this object-centric world model. Building on the efficient MBRL algorithm STORM, we call this pipeline OC-STORM. We demonstrate OC-STORM's practical value in overcoming the limitations of conventional MBRL approaches on both Atari games and the visually complex game Hollow Knight.

Trevor McInroe

We introduce Terra Nova, a new comprehensive challenge environment (CCE) for reinforcement learning (RL) research inspired by Civilization V. A CCE is a single environment in which multiple canonical RL challenges (e.g., partial observability, credit assignment, representation learning, enormous action spaces, etc.) arise simultaneously. Mastery therefore demands integrated, long-horizon understanding across many interacting variables. We emphasize that this definition excludes challenges that only aggregate unrelated tasks in independent, parallel streams (e.g., learning to play all Atari games at once). These aggregated multitask benchmarks primarily asses whether an agent can catalog and switch among unrelated policies rather than test an agent's ability to perform deep reasoning across many interacting challenges.

Elle Miller, Trevor McInroe, David Abel et al.

Achieving safe, reliable real-world robotic manipulation requires agents to evolve beyond vision and incorporate tactile sensing to overcome sensory deficits and reliance on idealised state information. Despite its potential, the efficacy of tactile sensing in reinforcement learning (RL) remains inconsistent. We address this by developing self-supervised learning (SSL) methodologies to more effectively harness tactile observations, focusing on a scalable setup of proprioception and sparse binary contacts. We empirically demonstrate that sparse binary tactile signals are critical for dexterity, particularly for interactions that proprioceptive control errors do not register, such as decoupled robot-object motions. Our agents achieve superhuman dexterity in complex contact tasks (ball bouncing and Baoding ball rotation). Furthermore, we find that decoupling the SSL memory from the on-policy memory can improve performance. We release the Robot Tactile Olympiad (RoTO) benchmark to standardise and promote future research in tactile-based manipulation. Project page: https://elle-miller.github.io/tactile_rl

Mhairi Dunion, Trevor McInroe, Kevin Sebastian Luck et al.

Reinforcement Learning (RL) environments can produce training data with spurious correlations between features due to the amount of training data or its limited feature coverage. This can lead to RL agents encoding these misleading correlations in their latent representation, preventing the agent from generalising if the correlation changes within the environment or when deployed in the real world. Disentangled representations can improve robustness, but existing disentanglement techniques that minimise mutual information between features require independent features, thus they cannot disentangle correlated features. We propose an auxiliary task for RL algorithms that learns a disentangled representation of high-dimensional observations with correlated features by minimising the conditional mutual information between features in the representation. We demonstrate experimentally, using continuous control tasks, that our approach improves generalisation under correlation shifts, as well as improving the training performance of RL algorithms in the presence of correlated features.

Trevor McInroe, Lukas Schäfer, Stefano V. Albrecht

Deep reinforcement learning (RL) agents that exist in high-dimensional state spaces, such as those composed of images, have interconnected learning burdens. Agents must learn an action-selection policy that completes their given task, which requires them to learn a representation of the state space that discerns between useful and useless information. The reward function is the only supervised feedback that RL agents receive, which causes a representation learning bottleneck that can manifest in poor sample efficiency. We present $k$-Step Latent (KSL), a new representation learning method that enforces temporal consistency of representations via a self-supervised auxiliary task wherein agents learn to recurrently predict action-conditioned representations of the state space. The state encoder learned by KSL produces low-dimensional representations that make optimization of the RL task more sample efficient. Altogether, KSL produces state-of-the-art results in both data efficiency and asymptotic performance in the popular PlaNet benchmark suite. Our analyses show that KSL produces encoders that generalize better to new tasks unseen during training, and its representations are more strongly tied to reward, are more invariant to perturbations in the state space, and move more smoothly through the temporal axis of the RL problem than other methods such as DrQ, RAD, CURL, and SAC-AE.