Manfred Eppe

Manfred Eppe, Christian Gumbsch, Matthias Kerzel et al.

According to cognitive psychology and related disciplines, the development of complex problem-solving behaviour in biological agents depends on hierarchical cognitive mechanisms. Hierarchical reinforcement learning is a promising computational approach that may eventually yield comparable problem-solving behaviour in artificial agents and robots. However, to date the problem-solving abilities of many human and non-human animals are clearly superior to those of artificial systems. Here, we propose steps to integrate biologically inspired hierarchical mechanisms to enable advanced problem-solving skills in artificial agents. Therefore, we first review the literature in cognitive psychology to highlight the importance of compositional abstraction and predictive processing. Then we relate the gained insights with contemporary hierarchical reinforcement learning methods. Interestingly, our results suggest that all identified cognitive mechanisms have been implemented individually in isolated computational architectures, raising the question of why there exists no single unifying architecture that integrates them. As our final contribution, we address this question by providing an integrative perspective on the computational challenges to develop such a unifying architecture. We expect our results to guide the development of more sophisticated cognitively inspired hierarchical machine learning architectures.

Frank Röder, Manfred Eppe, Stefan Wermter

This paper focuses on robotic reinforcement learning with sparse rewards for natural language goal representations. An open problem is the sample-inefficiency that stems from the compositionality of natural language, and from the grounding of language in sensory data and actions. We address these issues with three contributions. We first present a mechanism for hindsight instruction replay utilizing expert feedback. Second, we propose a seq2seq model to generate linguistic hindsight instructions. Finally, we present a novel class of language-focused learning tasks. We show that hindsight instructions improve the learning performance, as expected. In addition, we also provide an unexpected result: We show that the learning performance of our agent can be improved by one third if, in a sense, the agent learns to talk to itself in a self-supervised manner. We achieve this by learning to generate linguistic instructions that would have been appropriate as a natural language goal for an originally unintended behavior. Our results indicate that the performance gain increases with the task-complexity.

Frank Röder, Manfred Eppe

Human-to-human conversation is not just talking and listening. It is an incremental process where participants continually establish a common understanding to rule out misunderstandings. Current language understanding methods for intelligent robots do not consider this. There exist numerous approaches considering non-understandings, but they ignore the incremental process of resolving misunderstandings. In this article, we present a first formalization and experimental validation of incremental action-repair for robotic instruction-following based on reinforcement learning. To evaluate our approach, we propose a collection of benchmark environments for action correction in language-conditioned reinforcement learning, utilizing a synthetic instructor to generate language goals and their corresponding corrections. We show that a reinforcement learning agent can successfully learn to understand incremental corrections of misunderstood instructions.

Jan Benad, Pradeep Kr. Banerjee, Frank Röder et al.

Zero-shot generalization in contextual reinforcement learning remains a core challenge, particularly when the context is latent and must be inferred from data. A canonical failure mode is actuator inversion, where identical actions produce opposite physical effects under a latent binary context. We propose DMA*-SH, a framework where a single hypernetwork, trained solely via dynamics prediction, generates a small set of adapter weights shared across the dynamics model, policy, and action-value function. This shared modulation imparts an inductive bias matched to actuator inversion, while input/output normalization and random input masking stabilize context inference, promoting directionally concentrated representations. We provide theoretical support via an expressivity separation result for hypernetwork modulation, and a variance decomposition with policy-gradient variance bounds that formalize how within-mode compression improves learning under actuator inversion. For evaluation, we introduce the Actuator Inversion Benchmark (AIB), a suite of environments designed to isolate discontinuous context-to-dynamics interactions. On AIB's held-out actuator-inversion tasks, DMA*-SH achieves zero-shot generalization, outperforming domain randomization by 111.8% and surpassing a standard context-aware baseline by 16.1%.

Jan Dohmen, Frank Röder, Manfred Eppe

One problem with researching cognitive modeling and reinforcement learning (RL) is that researchers spend too much time on setting up an appropriate computational framework for their experiments. Many open source implementations of current RL algorithms exist, but there is a lack of a modular suite of tools combining different robotic simulators and platforms, data visualization, hyperparameter optimization, and baseline experiments. To address this problem, we present Scilab-RL, a software framework for efficient research in cognitive modeling and reinforcement learning for robotic agents. The framework focuses on goal-conditioned reinforcement learning using Stable Baselines 3 and the OpenAI gym interface. It enables native possibilities for experiment visualizations and hyperparameter optimization. We describe how these features enable researchers to conduct experiments with minimal time effort, thus maximizing research output.

Frank Röder, Jan Benad, Manfred Eppe et al.

Real-world reinforcement learning demands adaptation to unseen environmental conditions without costly retraining. Contextual Markov Decision Processes (cMDP) model this challenge, but existing methods often require explicit context variables (e.g., friction, gravity), limiting their use when contexts are latent or hard to measure. We introduce Dynamics-Aligned Latent Imagination (DALI), a framework integrated within the Dreamer architecture that infers latent context representations from agent-environment interactions. By training a self-supervised encoder to predict forward dynamics, DALI generates actionable representations conditioning the world model and policy, bridging perception and control. We theoretically prove this encoder is essential for efficient context inference and robust generalization. DALI's latent space enables counterfactual consistency: Perturbing a gravity-encoding dimension alters imagined rollouts in physically plausible ways. On challenging cMDP benchmarks, DALI achieves significant gains over context-unaware baselines, often surpassing context-aware baselines in extrapolation tasks, enabling zero-shot generalization to unseen contextual variations.

Sascha Griffiths, Tayfun Alpay, Alexander Sutherland et al.

In this paper, we investigate the roles that social robots can take in physical exercise with human partners. In related work, robots or virtual intelligent agents take the role of a coach or instructor whereas in other approaches they are used as motivational aids. These are two "paradigms", so to speak, within the small but growing area of robots for social exercise. We designed an online questionnaire to test whether the preferred role in which people want to see robots would be the companion or the coach. The questionnaire asks people to imagine working out with a robot with the help of three utilized questionnaires: (1) CART-Q which is used for judging coach-athlete relationships, (2) the mind perception questionnaire and (3) the System Usability Scale (SUS). We present the methodology, some preliminary results as well as our intended future work on personal robots for coaching.

Manfred Eppe, Pierre-Yves Oudeyer

This paper outlines a perspective on the future of AI, discussing directions for machines models of human-like intelligence. We explain how developmental and evolutionary theories of human cognition should further inform artificial intelligence. We emphasize the role of ecological niches in sculpting intelligent behavior, and in particular that human intelligence was fundamentally shaped to adapt to a constantly changing socio-cultural environment. We argue that a major limit of current work in AI is that it is missing this perspective, both theoretically and experimentally. Finally, we discuss the promising approach of developmental artificial intelligence, modeling infant development through multi-scale interaction between intrinsically motivated learning, embodiment and a fastly changing socio-cultural environment. This paper takes the form of an interview of Pierre-Yves Oudeyer by Mandred Eppe, organized within the context of a KI - K{ü}nstliche Intelligenz special issue in developmental robotics.

Manfred Eppe, Christian Gumbsch, Matthias Kerzel et al.

Cognitive Psychology and related disciplines have identified several critical mechanisms that enable intelligent biological agents to learn to solve complex problems. There exists pressing evidence that the cognitive mechanisms that enable problem-solving skills in these species build on hierarchical mental representations. Among the most promising computational approaches to provide comparable learning-based problem-solving abilities for artificial agents and robots is hierarchical reinforcement learning. However, so far the existing computational approaches have not been able to equip artificial agents with problem-solving abilities that are comparable to intelligent animals, including human and non-human primates, crows, or octopuses. Here, we first survey the literature in Cognitive Psychology, and related disciplines, and find that many important mental mechanisms involve compositional abstraction, curiosity, and forward models. We then relate these insights with contemporary hierarchical reinforcement learning methods, and identify the key machine intelligence approaches that realise these mechanisms. As our main result, we show that all important cognitive mechanisms have been implemented independently in isolated computational architectures, and there is simply a lack of approaches that integrate them appropriately. We expect our results to guide the development of more sophisticated cognitively inspired hierarchical methods, so that future artificial agents achieve a problem-solving performance on the level of intelligent animals.

Phuong D. H. Nguyen, Yasmin Kim Georgie, Ezgi Kayhan et al.

Safe human-robot interactions require robots to be able to learn how to behave appropriately in \sout{humans' world} \rev{spaces populated by people} and thus to cope with the challenges posed by our dynamic and unstructured environment, rather than being provided a rigid set of rules for operations. In humans, these capabilities are thought to be related to our ability to perceive our body in space, sensing the location of our limbs during movement, being aware of other objects and agents, and controlling our body parts to interact with them intentionally. Toward the next generation of robots with bio-inspired capacities, in this paper, we first review the developmental processes of underlying mechanisms of these abilities: The sensory representations of body schema, peripersonal space, and the active self in humans. Second, we provide a survey of robotics models of these sensory representations and robotics models of the self; and we compare these models with the human counterparts. Finally, we analyse what is missing from these robotics models and propose a theoretical computational framework, which aims to allow the emergence of the sense of self in artificial agents by developing sensory representations through self-exploration.

Phuong D. H. Nguyen, Manfred Eppe, Stefan Wermter

Cognitive science suggests that the self-representation is critical for learning and problem-solving. However, there is a lack of computational methods that relate this claim to cognitively plausible robots and reinforcement learning. In this paper, we bridge this gap by developing a model that learns bidirectional action-effect associations to encode the representations of body schema and the peripersonal space from multisensory information, which is named multimodal BidAL. Through three different robotic experiments, we demonstrate that this approach significantly stabilizes the learning-based problem-solving under noisy conditions and that it improves transfer learning of robotic manipulation skills.

Thilo Fryen, Manfred Eppe, Phuong D. H. Nguyen et al.

Reinforcement learning is a promising method to accomplish robotic control tasks. The task of playing musical instruments is, however, largely unexplored because it involves the challenge of achieving sequential goals - melodies - that have a temporal dimension. In this paper, we address robotic musicianship by introducing a temporal extension to goal-conditioned reinforcement learning: Time-dependent goals. We demonstrate that these can be used to train a robotic musician to play the theremin instrument. We train the robotic agent in simulation and transfer the acquired policy to a real-world robotic thereminist. Supplemental video: https://youtu.be/jvC9mPzdQN4

Matthias Kerzel, Fares Abawi, Manfred Eppe et al.

We present a follow-up study on our unified visuomotor neural model for the robotic tasks of identifying, localizing, and grasping a target object in a scene with multiple objects. Our Retinanet-based model enables end-to-end training of visuomotor abilities in a biologically inspired developmental approach. In our initial implementation, a neural model was able to grasp selected objects from a planar surface. We embodied the model on the NICO humanoid robot. In this follow-up study, we expand the task and the model to reaching for objects in a three-dimensional space with a novel dataset based on augmented reality and a simulation environment. We evaluate the influence of training with auxiliary tasks, i.e., if learning of the primary visuomotor task is supported by learning to classify and locate different objects. We show that the proposed visuomotor model can learn to reach for objects in a three-dimensional space. We analyze the results for biologically-plausible biases based on object locations or properties. We show that the primary visuomotor task can be successfully trained simultaneously with one of the two auxiliary tasks. This is enabled by a complex neurocognitive model with shared and task-specific components, similar to models found in biological systems.

Frank Röder, Manfred Eppe, Phuong D. H. Nguyen et al.

Hierarchical abstraction and curiosity-driven exploration are two common paradigms in current reinforcement learning approaches to break down difficult problems into a sequence of simpler ones and to overcome reward sparsity. However, there is a lack of approaches that combine these paradigms, and it is currently unknown whether curiosity also helps to perform the hierarchical abstraction. As a novelty and scientific contribution, we tackle this issue and develop a method that combines hierarchical reinforcement learning with curiosity. Herein, we extend a contemporary hierarchical actor-critic approach with a forward model to develop a hierarchical notion of curiosity. We demonstrate in several continuous-space environments that curiosity can more than double the learning performance and success rates for most of the investigated benchmarking problems. We also provide our source code and a supplementary video.

Manfred Eppe, Phuong D. H. Nguyen, Stefan Wermter

Reinforcement learning is an appropriate and successful method to robustly perform low-level robot control under noisy conditions. Symbolic action planning is useful to resolve causal dependencies and to break a causally complex problem down into a sequence of simpler high-level actions. A problem with the integration of both approaches is that action planning is based on discrete high-level action- and state spaces, whereas reinforcement learning is usually driven by a continuous reward function. However, recent advances in reinforcement learning, specifically, universal value function approximators and hindsight experience replay, have focused on goal-independent methods based on sparse rewards. In this article, we build on these novel methods to facilitate the integration of action planning with reinforcement learning by exploiting the reward-sparsity as a bridge between the high-level and low-level state- and control spaces. As a result, we demonstrate that the integrated neuro-symbolic method is able to solve object manipulation problems that involve tool use and non-trivial causal dependencies under noisy conditions, exploiting both data and knowledge.

Manfred Eppe, Sven Magg, Stefan Wermter

Deep reinforcement learning has recently gained a focus on problems where policy or value functions are independent of goals. Evidence exists that the sampling of goals has a strong effect on the learning performance, but there is a lack of general mechanisms that focus on optimizing the goal sampling process. In this work, we present a simple and general goal masking method that also allows us to estimate a goal's difficulty level and thus realize a curriculum learning approach for deep RL. Our results indicate that focusing on goals with a medium difficulty level is appropriate for deep deterministic policy gradient (DDPG) methods, while an "aim for the stars and reach the moon-strategy", where hard goals are sampled much more often than simple goals, leads to the best learning performance in cases where DDPG is combined with for hindsight experience replay (HER). We demonstrate that the approach significantly outperforms standard goal sampling for different robotic object manipulation problems.

Manfred Eppe, Matthias Kerzel, Erik Strahl et al.

We present a novel approach for interactive auditory object analysis with a humanoid robot. The robot elicits sensory information by physically shaking visually indistinguishable plastic capsules. It gathers the resulting audio signals from microphones that are embedded into the robotic ears. A neural network architecture learns from these signals to analyze properties of the contents of the containers. Specifically, we evaluate the material classification and weight prediction accuracy and demonstrate that the framework is fairly robust to acoustic real-world noise.

Michael Spranger, Jakob Suchan, Mehul Bhatt et al.

This paper presents a computational model of the processing of dynamic spatial relations occurring in an embodied robotic interaction setup. A complete system is introduced that allows autonomous robots to produce and interpret dynamic spatial phrases (in English) given an environment of moving objects. The model unites two separate research strands: computational cognitive semantics and on commonsense spatial representation and reasoning. The model for the first time demonstrates an integration of these different strands.

Manfred Eppe, Sean Trott, Jerome Feldman

We develop a natural language interface for human robot interaction that implements reasoning about deep semantics in natural language. To realize the required deep analysis, we employ methods from cognitive linguistics, namely the modular and compositional framework of Embodied Construction Grammar (ECG) [Feldman, 2009]. Using ECG, robots are able to solve fine-grained reference resolution problems and other issues related to deep semantics and compositionality of natural language. This also includes verbal interaction with humans to clarify commands and queries that are too ambiguous to be executed safely. We implement our NLU framework as a ROS package and present proof-of-concept scenarios with different robots, as well as a survey on the state of the art.

Manfred Eppe

We present an epistemic action theory for tractable epistemic reasoning as an extension to the h-approximation (HPX) theory. In contrast to existing tractable approaches, the theory supports functional fluents and postdictive reasoning with static causal laws. We argue that this combination is particularly synergistic because it allows one not only to perform direct postdiction about the conditions of actions, but also indirect postdiction about the conditions of static causal laws. We show that despite the richer expressiveness, the temporal projection problem remains tractable (polynomial), and therefore the planning problem remains in NP. We present the operational semantics of our theory as well as its formulation as Answer Set Programming.

Manfred Eppe, Mehul Bhatt

Making sense of incomplete and conflicting narrative knowledge in the presence of abnormalities, unobservable processes, and other real world considerations is a challenge and crucial requirement for cognitive robotics systems. An added challenge, even when suitably specialised action languages and reasoning systems exist, is practical integration and application within large-scale robot control frameworks. In the backdrop of an autonomous wheelchair robot control task, we report on application-driven work to realise postdiction triggered abnormality detection and re-planning for real-time robot control: (a) Narrative-based knowledge about the environment is obtained via a larger smart environment framework; and (b) abnormalities are postdicted from stable-models of an answer-set program corresponding to the robot's epistemic model. The overall reasoning is performed in the context of an approximate epistemic action theory based planner implemented via a translation to answer-set programming.

Manfred Eppe, Mehul Bhatt, Frank Dylla

We propose an approximation of the Possible Worlds Semantics (PWS) for action planning. A corresponding planning system is implemented by a transformation of the action specification to an Answer-Set Program. A novelty is support for postdiction wrt. (a) the plan existence problem in our framework can be solved in NP, as compared to $Σ_2^P$ for non-approximated PWS of Baral(2000); and (b) the planner generates optimal plans wrt. a minimal number of actions in $Δ_2^P$. We demo the planning system with standard problems, and illustrate its integration in a larger software framework for robot control in a smart home.