arXiv subjects

Lukas Schäfer

Publications and source records attributed to Lukas Schäfer.

At least 19 recordsLinked to original sources

arXiv2026

Augmentations for Robust and Efficient Imitation Learning in Streamed Video Games

Imitation learning is an appealing way to scale game-playing agents to complex 3D environments by training policies to map visual observations to actions from human demonstrations. However, these demonstrations are expensive to collect and modern game-playing is often done through streaming in which network delay and compression introduce spatiotemporally correlated visual artifacts that can cause a covariance shift at test time. To address these challenges, we propose streaming augmentations that mimic four types of artifacts commonly encountered during streaming with low-bandwidth network connection: pixelated blocks and scrubs, global blur, and ghosting. We instantiate our approach on top of predictive inverse dynamics models (PIDM), which combine future-state conditioning with an inverse dynamics policy in a learned latent space, and evaluate the impact of our augmentations across three tasks in modern 3D video games. Under stable streaming conditions, agents trained with spatiotemporal augmentations achieve up to 41% higher evaluation performance compared to agents trained without augmentations under an identical data budget. When network lag is introduced, agents trained with augmentations degrade by only 7.45% vs 49.82% of the original performance for agents trained only with the original data. These results clearly indicate that spatiotemporal augmentations tailored for the streaming setting are a simple yet powerful tool to train robust and efficient game-playing agents.

cs.LG↗

arXiv2026

When Does Predictive Inverse Dynamics Outperform Behavior Cloning?

Behavior cloning (BC) is a practical offline imitation learning method, but it often fails when expert demonstrations are limited. Recent works have introduced a class of architectures named predictive inverse dynamics models (PIDMs) that combine a future-state predictor with an inverse dynamics model. While PIDMs often outperform BC, the reasons behind their benefits remain unclear. In this paper, we provide a theoretical explanation: PIDMs introduce a tradeoff. Conditioning the IDM on the predicted future state can significantly reduce variance, but the prediction itself introduces additional bias and variance. We establish conditions for PIDMs to achieve higher sample efficiency and lower prediction error than BC, with the gap widening when additional data sources are available. We validate the theoretical insights empirically in 2D navigation tasks, where BC requires up to five times (three times on average) more demonstrations than PIDM to reach comparable performance. Results are also illustrated in a complex 3D environment in a modern video game with high-dimensional visual inputs and stochastic transitions, where BC requires over 66\% more samples than PIDM.

cs.LG↗

arXiv2026

Safe Reinforcement Learning using Action Projection: Safeguard the Policy or the Environment?

Projection-based safety filters, which modify unsafe actions by mapping them to the closest safe alternative, are widely used to enforce safety constraints in reinforcement learning (RL). Two integration strategies are commonly considered: Safe environment RL (SE-RL), where the safeguard is treated as part of the environment, and safe policy RL (SP-RL), where it is embedded within the policy through differentiable optimization layers. Despite their practical relevance in safety-critical settings, a formal understanding of their differences is lacking. In this work, we present a theoretical comparison of SE-RL and SP-RL. We identify a key distinction in how each approach is affected by action aliasing, a phenomenon in which multiple unsafe actions are projected to the same safe action, causing information loss in the policy gradients. In SE-RL, this effect is implicitly approximated by the critic, while in SP-RL, it manifests directly as rank-deficient Jacobians during backpropagation through the safeguard. Our contributions are threefold: (i) a unified formalization of SE-RL and SP-RL in the context of actor-critic algorithms, (ii) a theoretical analysis of their respective policy gradient estimates, highlighting the role of action aliasing, and (iii) a comparative study of mitigation strategies, including a novel penalty-based improvement for SP-RL that aligns with established SE-RL practices. Empirical results support our theoretical predictions, showing that action aliasing is more detrimental for SP-RL than for SE-RL. However, with appropriate improvement strategies, SP-RL can match or outperform improved SE-RL across a range of environments. These findings provide actionable insights for choosing and refining projection-based safe RL methods based on task characteristics.

cs.LG↗

arXiv2025

From Zonotopes to Proof Certificates: A Formal Pipeline for Safe Control Envelopes

Synthesizing controllers that enforce both safety and actuator constraints is a central challenge in the design of cyber-physical systems. State-of-the-art reachability methods based on zonotopes deliver impressive scalability, yet no zonotope reachability tool has been formally verified and the lack of end-to-end correctness undermines the confidence in their use for safety-critical systems. Although deductive verification with the hybrid system prover KeYmaera X could, in principle, resolve this assurance gap, the high-dimensional set representations required for realistic control envelopes overwhelm its reasoning based on quantifier elimination. To address this gap, we formalize how control-invariant sets serve as sound safety certificates. Building on that foundation, we develop a verification pipeline for control envelopes that unites scalability and formal rigor. First, we compute control envelopes with high-performance reachability algorithms. Second, we certify every intermediate result using provably correct logical principles. To accelerate this certification, we offload computationally intensive zonotope containment tasks to efficient numerical backends, which return compact witnesses that KeYmaera X validates rapidly. We show the practical utility of our approach through representative case studies.

cs.LO↗

arXiv2025

Approximability of the Containment Problem for Zonotopes and Ellipsotopes

The zonotope containment problem, i.e., whether one zonotope is contained in another, is a central problem in control theory. Applications include detecting faults and robustifying controllers by computing invariant sets, and obtain fixed points in reachability analysis. Despite the inherent co-NP-hardness of this problem, an approximation algorithm developed by S. Sadraddini and R. Tedrake has gained widespread recognition for its swift execution and consistent reliability in practice. In our study, we substantiate the precision of the algorithm with a definitive proof, elucidating the empirical accuracy observed in practice. Our proof hinges on establishing a connection between the containment problem and the computation of matrix norms, thereby enabling the extension of the approximation algorithm to encompass ellipsotopes -- a broader class of sets derived from zonotopes. We also explore the computational complexity of the ellipsotope containment problem with a focus on approximability. Finally, we present new methods to compute safe sets for linear dynamical systems, demonstrating the practical relevance of approximating the ellipsotope containment problem.

math.OC↗

arXiv2025

Visual Encoders for Data-Efficient Imitation Learning in Modern Video Games

Video games have served as useful benchmarks for the decision-making community, but going beyond Atari games towards modern games has been prohibitively expensive for the vast majority of the research community. Prior work in modern video games typically relied on game-specific integration to obtain game features and enable online training, or on existing large datasets. An alternative approach is to train agents using imitation learning to play video games purely from images. However, this setting poses a fundamental question: which visual encoders obtain representations that retain information critical for decision making? To answer this question, we conduct a systematic study of imitation learning with publicly available pre-trained visual encoders compared to the typical task-specific end-to-end training approach in Minecraft, Counter-Strike: Global Offensive, and Minecraft Dungeons. Our results show that end-to-end training can be effective with comparably low-resolution images and only minutes of demonstrations, but significant improvements can be gained by utilising pre-trained encoders such as DINOv2 depending on the game. In addition to enabling effective decision making, we show that pre-trained encoders can make decision-making research in video games more accessible by significantly reducing the cost of training.

cs.LG↗

arXiv2025

Ensemble Value Functions for Efficient Exploration in Multi-Agent Reinforcement Learning

Multi-agent reinforcement learning (MARL) requires agents to explore within a vast joint action space to find joint actions that lead to coordination. Existing value-based MARL algorithms commonly rely on random exploration, such as $ε$-greedy, to explore the environment which is not systematic and inefficient at identifying effective actions in multi-agent problems. Additionally, the concurrent training of the policies of multiple agents during training can render the optimisation non-stationary. This can lead to unstable value estimates, highly variant gradients, and ultimately hinder coordination between agents. To address these challenges, we propose ensemble value functions for multi-agent exploration (EMAX). EMAX is a framework to seamlessly extend value-based MARL algorithms. EMAX leverages an ensemble of value functions for each agent to guide their exploration, reduce the variance of their optimisation, and makes their policies more robust to miscoordination. EMAX achieves these benefits by (1) systematically guiding the exploration of agents with a UCB policy towards parts of the environment that require multiple agents to coordinate. (2) EMAX computes average value estimates across the ensemble as target values to reduce the variance of gradients and make optimisation more stable. (3) During evaluation, EMAX selects actions following a majority vote across the ensemble to reduce the likelihood of miscoordination. We first instantiate independent DQN with EMAX and evaluate it in 11 general-sum tasks with sparse rewards. We show that EMAX improves final evaluation returns by 185% across all tasks. We then evaluate EMAX on top of IDQN, VDN and QMIX in 21 common-reward tasks, and show that EMAX improves sample efficiency and final evaluation returns across all tasks over all three vanilla algorithms by 60%, 47%, and 538%, respectively.

cs.MA↗

arXiv2024

Using Offline Data to Speed Up Reinforcement Learning in Procedurally Generated Environments

One of the key challenges of Reinforcement Learning (RL) is the ability of agents to generalise their learned policy to unseen settings. Moreover, training RL agents requires large numbers of interactions with the environment. Motivated by the recent success of Offline RL and Imitation Learning (IL), we conduct a study to investigate whether agents can leverage offline data in the form of trajectories to improve the sample-efficiency in procedurally generated environments. We consider two settings of using IL from offline data for RL: (1) pre-training a policy before online RL training and (2) concurrently training a policy with online RL and IL from offline data. We analyse the impact of the quality (optimality of trajectories) and diversity (number of trajectories and covered level) of available offline trajectories on the effectiveness of both approaches. Across four well-known sparse reward tasks in the MiniGrid environment, we find that using IL for pre-training and concurrently during online RL training both consistently improve the sample-efficiency while converging to optimal policies. Furthermore, we show that pre-training a policy from as few as two trajectories can make the difference between learning an optimal policy at the end of online training and not learning at all. Our findings motivate the widespread adoption of IL for pre-training and concurrent IL in procedurally generated environments whenever offline trajectories are available or can be generated.

cs.LG↗

arXiv2024

Scalable Multi-Agent Reinforcement Learning for Warehouse Logistics with Robotic and Human Co-Workers

We consider a warehouse in which dozens of mobile robots and human pickers work together to collect and deliver items within the warehouse. The fundamental problem we tackle, called the order-picking problem, is how these worker agents must coordinate their movement and actions in the warehouse to maximise performance in this task. Established industry methods using heuristic approaches require large engineering efforts to optimise for innately variable warehouse configurations. In contrast, multi-agent reinforcement learning (MARL) can be flexibly applied to diverse warehouse configurations (e.g. size, layout, number/types of workers, item replenishment frequency), and different types of order-picking paradigms (e.g. Goods-to-Person and Person-to-Goods), as the agents can learn how to cooperate optimally through experience. We develop hierarchical MARL algorithms in which a manager agent assigns goals to worker agents, and the policies of the manager and workers are co-trained toward maximising a global objective (e.g. pick rate). Our hierarchical algorithms achieve significant gains in sample efficiency over baseline MARL algorithms and overall pick rates over multiple established industry heuristics in a diverse set of warehouse configurations and different order-picking paradigms.

cs.LG↗

arXiv2024

Multi-Horizon Representations with Hierarchical Forward Models for Reinforcement Learning

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.

cs.LG↗

arXiv2023

Magneto-active composites with locally tailored stiffness produced by laser powder bed fusion

Additive manufacturing technologies enable the production of complex and bioinspired shapes using magneto-responsive materials, which find diverse applications in soft robotics. Particularly, the development of composites with controlled gradients in mechanical properties offers new prospects for advancements in magneto-active materials. However, achieving such composites with gradients typically involves complex multi-material printing procedures. In this study, a single-step laser powder bed fusion (LPBF) process is proposed that enables precise local adjustments of the mechanical stiffness within magneto-active composites. By utilizing distinct laser parameters in specific regions of a composite containing thermoplastic polyurethane and atomized magnetic powder derived from hard magnetic Nd-Fe-B, the stiffness of the composite can be modified within the range of 2 to 22 MPa. Various magneto-responsive actuators with locally tailored stiffness are fabricated and their magnetic performance is investigated. The enhanced response exhibited by actuators with locally adjusted mechanical properties in comparison to their homogeneous counterparts with identical geometries is shown. As a demonstration of a biomedical application, a magnetically responsive stent with localized adjustment is presented with the ability to meet specific requirements in terms of geometry and local stiffness based on an individual's anatomy and disease condition. The proposed method presents an approach for creating functionally graded materials using LPBF, not only for magneto-active materials but also for several other structural and functional materials.

physics.app-ph↗

arXiv2023

Learning Task Embeddings for Teamwork Adaptation in Multi-Agent Reinforcement Learning

Successful deployment of multi-agent reinforcement learning often requires agents to adapt their behaviour. In this work, we discuss the problem of teamwork adaptation in which a team of agents needs to adapt their policies to solve novel tasks with limited fine-tuning. Motivated by the intuition that agents need to be able to identify and distinguish tasks in order to adapt their behaviour to the current task, we propose to learn multi-agent task embeddings (MATE). These task embeddings are trained using an encoder-decoder architecture optimised for reconstruction of the transition and reward functions which uniquely identify tasks. We show that a team of agents is able to adapt to novel tasks when provided with task embeddings. We propose three MATE training paradigms: independent MATE, centralised MATE, and mixed MATE which vary in the information used for the task encoding. We show that the embeddings learned by MATE identify tasks and provide useful information which agents leverage during adaptation to novel tasks.

cs.MA↗

arXiv2023

Provably Safe Reinforcement Learning: Conceptual Analysis, Survey, and Benchmarking

Ensuring the safety of reinforcement learning (RL) algorithms is crucial to unlock their potential for many real-world tasks. However, vanilla RL and most safe RL approaches do not guarantee safety. In recent years, several methods have been proposed to provide hard safety guarantees for RL, which is essential for applications where unsafe actions could have disastrous consequences. Nevertheless, there is no comprehensive comparison of these provably safe RL methods. Therefore, we introduce a categorization of existing provably safe RL methods, present the conceptual foundations for both continuous and discrete action spaces, and empirically benchmark existing methods. We categorize the methods based on how they adapt the action: action replacement, action projection, and action masking. Our experiments on an inverted pendulum and a quadrotor stabilization task indicate that action replacement is the best-performing approach for these applications despite its comparatively simple realization. Furthermore, adding a reward penalty, every time the safety verification is engaged, improved training performance in our experiments. Finally, we provide practical guidance on selecting provably safe RL approaches depending on the safety specification, RL algorithm, and type of action space.

cs.LG↗

arXiv2023

Magnetic properties of Nd6Fe13Cu single crystals

The understanding of coercivity mechanism in high performance Nd-Fe-B permanent magnets relies on the analysis of the magnetic properties of all phases present in the magnets. By adding Cu in such compounds, a new Nd6Fe13Cu grain boundary phase is formed, however, the magnetic properties of this phase and its role in the magnetic decoupling of the matrix Nd2Fe14B grains are still insufficiently studied. In this work, we have grown Nd6Fe13Cu single crystals by the reactive flux method and studied their magnetic properties in detail. It is observed that below the Néel temperature (TN = 410 K), the Nd6Fe13Cu is antiferromagnetic in zero magnetic field; whereas when a magnetic field is applied along the a-axis, a spin-flop transition occurs at approx. 6 T, indicating a strong competition between antiferromagnetic and ferromagnetic interactions in two Nd layers below and above the Cu layers. Our atomistic spin dynamics simulation confirms that an increase in temperature and/or magnetic field can significantly change the antiferromagnetic coupling between the two Nd layers below and above the Cu layers, which, in turn, is the reason for the observed spin-flop transition. These results suggest that the role of antiferromagnetic Nd6Fe13Cu grain boundary phase in the coercivity enhancement of Nd-Fe-B-Cu magnets is more complex than previously thought, mainly due to the competition between its antiferro- and ferro-magnetic exchange interactions.

cond-mat.mtrl-sci↗

arXiv2022

Robust On-Policy Sampling for Data-Efficient Policy Evaluation in Reinforcement Learning

Reinforcement learning (RL) algorithms are often categorized as either on-policy or off-policy depending on whether they use data from a target policy of interest or from a different behavior policy. In this paper, we study a subtle distinction between on-policy data and on-policy sampling in the context of the RL sub-problem of policy evaluation. We observe that on-policy sampling may fail to match the expected distribution of on-policy data after observing only a finite number of trajectories and this failure hinders data-efficient policy evaluation. Towards improved data-efficiency, we show how non-i.i.d., off-policy sampling can produce data that more closely matches the expected on-policy data distribution and consequently increases the accuracy of the Monte Carlo estimator for policy evaluation. We introduce a method called Robust On-Policy Sampling and demonstrate theoretically and empirically that it produces data that converges faster to the expected on-policy distribution compared to on-policy sampling. Empirically, we show that this faster convergence leads to lower mean squared error policy value estimates.

cs.LG↗

arXiv2022

Deep Reinforcement Learning for Multi-Agent Interaction

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.

cs.MA↗

arXiv2022

Decoupled Reinforcement Learning to Stabilise Intrinsically-Motivated Exploration

Intrinsic rewards can improve exploration in reinforcement learning, but the exploration process may suffer from instability caused by non-stationary reward shaping and strong dependency on hyperparameters. In this work, we introduce Decoupled RL (DeRL) as a general framework which trains separate policies for intrinsically-motivated exploration and exploitation. Such decoupling allows DeRL to leverage the benefits of intrinsic rewards for exploration while demonstrating improved robustness and sample efficiency. We evaluate DeRL algorithms in two sparse-reward environments with multiple types of intrinsic rewards. Our results show that DeRL is more robust to varying scale and rate of decay of intrinsic rewards and converges to the same evaluation returns than intrinsically-motivated baselines in fewer interactions. Lastly, we discuss the challenge of distribution shift and show that divergence constraint regularisers can successfully minimise instability caused by divergence of exploration and exploitation policies.

cs.LG↗

arXiv2021

Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks

Multi-agent deep reinforcement learning (MARL) suffers from a lack of commonly-used evaluation tasks and criteria, making comparisons between approaches difficult. In this work, we provide a systematic evaluation and comparison of three different classes of MARL algorithms (independent learning, centralised multi-agent policy gradient, value decomposition) in a diverse range of cooperative multi-agent learning tasks. Our experiments serve as a reference for the expected performance of algorithms across different learning tasks, and we provide insights regarding the effectiveness of different learning approaches. We open-source EPyMARL, which extends the PyMARL codebase to include additional algorithms and allow for flexible configuration of algorithm implementation details such as parameter sharing. Finally, we open-source two environments for multi-agent research which focus on coordination under sparse rewards.

cs.LG↗