Jens Kober

Zhaoting Li, Gang Chen, Javier Alonso-Mora et al.

Diffusion policies have recently emerged as a powerful framework for robotic manipulation. However, like other behavior cloning methods, they remain vulnerable to distributional shift, often requiring human-in-the-loop interventions to correct failures during deployment. These interactions naturally provide paired supervision in the form of the robot's undesired actions and the human teacher's corrective actions. Yet existing data aggregation pipelines and standard behavior cloning losses largely ignore this negative signal from undesired actions, leading to overfitting to teacher's actions and an increasing reliance on costly expert data. To address this limitation, we propose Set-Supervised Diffusion Policy (SDP), a novel learning framework that utilizes contrastive action-chunk data to train diffusion policies from human corrections. From paired positive and negative action-chunks, SDP constructs a set of desired action-chunks and designs a training pipeline that encourages the diffusion policy to align with the set. Through extensive experiments across multiple robotic manipulation tasks, we demonstrate that SDP consistently improves policy performance, with particularly strong gains in robustness to noisy data. Moreover, SDP induces high-quality aggregated datasets, enabling more efficient and reliable policy learning from human-in-the-loop corrections. Our code is available at https://set-supervised-diffusion-policy.github.io/.

Julian F. Schumann, Johan Engström, Leif Johnson et al.

Collision avoidance -- involving a rapid threat detection and quick execution of the appropriate evasive maneuver -- is a critical aspect of driving. However, existing models of human collision avoidance behavior are fragmented, focusing on specific scenarios or only describing certain aspects of the avoidance behavior, such as response times. This paper addresses these gaps by proposing a novel computational cognitive model of human collision avoidance behavior based on active inference. Active inference provides a unified approach to modeling human behavior: the minimization of free energy. Building on prior active inference work, our model incorporates established cognitive mechanisms such as evidence accumulation to simulate human responses in two distinct collision avoidance scenarios: front-to-rear lead vehicle braking and lateral incursion by an oncoming vehicle. We demonstrate that our model explains a wide range of previous empirical findings on human collision avoidance behavior. Specifically, the model closely reproduces both aggregate results from meta-analyses previously reported in the literature and detailed, scenario-specific effects observed in a recent driving simulator study, including response timing, maneuver selection, and execution. Our results highlight the potential of active inference as a unified framework for understanding and modeling human behavior in complex real-life driving tasks.

Bas van der Heijden, Jelle Luijkx, Laura Ferranti et al.

Sim2real, that is, the transfer of learned control policies from simulation to real world, is an area of growing interest in robotics due to its potential to efficiently handle complex tasks. The sim2real approach faces challenges due to mismatches between simulation and reality. These discrepancies arise from inaccuracies in modeling physical phenomena and asynchronous control, among other factors. To this end, we introduce EAGERx, a framework with a unified software pipeline for both real and simulated robot learning. It can support various simulators and aids in integrating state, action and time-scale abstractions to facilitate learning. EAGERx's integrated delay simulation, domain randomization features, and proposed synchronization algorithm contribute to narrowing the sim2real gap. We demonstrate (in the context of robot learning and beyond) the efficacy of EAGERx in accommodating diverse robotic systems and maintaining consistent simulation behavior. EAGERx is open source and its code is available at https://eagerx.readthedocs.io.

Antonin Raffin, Daniel Seidel, Jens Kober et al.

Spring-based actuators in legged locomotion provide energy-efficiency and improved performance, but increase the difficulty of controller design. While previous work has focused on extensive modeling and simulation to find optimal controllers for such systems, we propose to learn model-free controllers directly on the real robot. In our approach, gaits are first synthesized by central pattern generators (CPGs), whose parameters are optimized to quickly obtain an open-loop controller that achieves efficient locomotion. Then, to make this controller more robust and further improve the performance, we use reinforcement learning to close the loop, to learn corrective actions on top of the CPGs. We evaluate the proposed approach on the DLR elastic quadruped bert. Our results in learning trotting and pronking gaits show that exploitation of the spring actuator dynamics emerges naturally from optimizing for dynamic motions, yielding high-performing locomotion, particularly the fastest walking gait recorded on bert, despite being model-free. The whole process takes no more than 1.5 hours on the real robot and results in natural-looking gaits.

Zhaoting Li, Rodrigo Pérez-Dattari, Robert Babuska et al.

Behavior cloning (BC) optimizes policies by treating human demonstrations as pointwise action labels. While effective with accurate action labels, this formulation is brittle in practice: when human-provided actions are imperfect, treating each label as an exact target can steer the policy away from the underlying desired behavior, particularly when expressive models are used (e.g., energy-based models). As a result, we propose a human-in-the-loop alternative that replaces pointwise supervision with set-valued action targets. We introduce Contrastive policy Learning from Interactive Corrections (CLIC). CLIC leverages human corrections to construct and refine sets of desired actions, and optimizes a policy to place probability mass over these sets rather than over a single action target. This formulation naturally accommodates both absolute and relative corrections and can represent complex multi-modal behaviors. Extensive simulation and real-robot experiments show that the proposed approach leads to effective policy learning across diverse settings: CLIC remains competitive with the state of the art under accurate data while being substantially more robust under noisy, relative, and partial feedback. Our implementation is publicly available at https://clic-webpage.github.io/.

Yulei Qiu, Jihong Zhu, Cosimo Della Santina et al.

Deformable Object Manipulation (DOM) is an important field of research as it contributes to practical tasks such as automatic cloth handling, cable routing, surgical operation, etc. Perception is considered one of the major challenges in DOM due to the complex dynamics and high degree of freedom of deformable objects. In this paper, we develop a novel image-processing algorithm based on Gabor filters to extract useful features from cloth, and based on this, devise a strategy for cloth flattening tasks. We also evaluate the overall framework experimentally and compare it with three human operators. The results show that our algorithm can determine the direction of wrinkles on the cloth accurately in simulation as well as in real robot experiments. Furthermore, our dewrinkling strategy compares favorably to baseline methods. The experiment video is available on https://sites.google.com/view/robotic-fabric-flattening/home

Julian Frederik Schumann, Jens Kober, Arkady Zgonnikov

Autonomous vehicles currently suffer from a time-inefficient driving style caused by uncertainty about human behavior in traffic interactions. Accurate and reliable prediction models enabling more efficient trajectory planning could make autonomous vehicles more assertive in such interactions. However, the evaluation of such models is commonly oversimplistic, ignoring the asymmetric importance of prediction errors and the heterogeneity of the datasets used for testing. We examine the potential of recasting interactions between vehicles as gap acceptance scenarios and evaluating models in this structured environment. To that end, we develop a framework aiming to facilitate the evaluation of any model, by any metric, and in any scenario. We then apply this framework to state-of-the-art prediction models, which all show themselves to be unreliable in the most safety-critical situations.

Yurui Du, Flavia Sofia Acerbo, Jens Kober et al.

In recent years, imitation learning (IL) has been widely used in industry as the core of autonomous vehicle (AV) planning modules. However, previous IL works show sample inefficiency and low generalisation in safety-critical scenarios, on which they are rarely tested. As a result, IL planners can reach a performance plateau where adding more training data ceases to improve the learnt policy. First, our work presents an IL model using the spline coefficient parameterisation and offline expert queries to enhance safety and training efficiency. Then, we expose the weakness of the learnt IL policy by synthetically generating critical scenarios through optimisation of parameters of the driver's risk field (DRF), a parametric human driving behaviour model implemented in a multi-agent traffic simulator based on the Lyft Prediction Dataset. To continuously improve the learnt policy, we retrain the IL model with augmented data. Thanks to the expressivity and interpretability of the DRF, the desired driving behaviours can be encoded and aggregated to the original training data. Our work constitutes a full development cycle that can efficiently and continuously improve the learnt IL policies in closed-loop. Finally, we show that our IL planner developed with less training resource still has superior performance compared to the previous state-of-the-art.

Jelle Luijkx, Zlatan Ajanovic, Laura Ferranti et al.

Several recent works show impressive results in mapping language-based human commands and image scene observations to direct robot executable policies (e.g., pick and place poses). However, these approaches do not consider the uncertainty of the trained policy and simply always execute actions suggested by the current policy as the most probable ones. This makes them vulnerable to domain shift and inefficient in the number of required demonstrations. We extend previous works and present the PARTNR algorithm that can detect ambiguities in the trained policy by analyzing multiple modalities in the pick and place poses using topological analysis. PARTNR employs an adaptive, sensitivity-based, gating function that decides if additional user demonstrations are required. User demonstrations are aggregated to the dataset and used for subsequent training. In this way, the policy can adapt promptly to domain shift and it can minimize the number of required demonstrations for a well-trained policy. The adaptive threshold enables to achieve the user-acceptable level of ambiguity to execute the policy autonomously and in turn, increase the trustworthiness of our system. We demonstrate the performance of PARTNR in a table-top pick and place task.

Eveline Drijver, Rodrigo Pérez-Dattari, Jens Kober et al.

Intelligent manufacturing is becoming increasingly important due to the growing demand for maximizing productivity and flexibility while minimizing waste and lead times. This work investigates automated secondary robotic food packaging solutions that transfer food products from the conveyor belt into containers. A major problem in these solutions is varying product supply which can cause drastic productivity drops. Conventional rule-based approaches, used to address this issue, are often inadequate, leading to violation of the industry's requirements. Reinforcement learning, on the other hand, has the potential of solving this problem by learning responsive and predictive policy, based on experience. However, it is challenging to utilize it in highly complex control schemes. In this paper, we propose a reinforcement learning framework, designed to optimize the conveyor belt speed while minimizing interference with the rest of the control system. When tested on real-world data, the framework exceeds the performance requirements (99.8% packed products) and maintains quality (100% filled boxes). Compared to the existing solution, our proposed framework improves productivity, has smoother control, and reduces computation time.

Julian F. Schumann, Johan Engström, Ran Wei et al.

Understanding how road users resolve space-sharing conflicts is important both for traffic safety and the safe deployment of autonomous vehicles. While existing models have captured specific aspects of such interactions (e.g., explicit communication), a theoretically-grounded computational framework has been lacking. In this paper, we extend a previously developed active inference-based driver behavior model to simulate interactive behavior of two agents. Our model captures three complementary mechanisms for uncertainty reduction in interaction: (i) implicit communication via direct behavioral coupling, (ii) reliance on normative expectations (stop signs, priority rules, etc.), and (iii) explicit communication. In a simplified intersection scenario, we show that normative and explicit communication cues can increase the likelihood of a successful conflict resolution. However, this relies on agents acting as expected. In situations where another agent (intentionally or unintentionally) violates normative expectations or communicates misleading information, reliance on these cues may induce collisions. These findings illustrate how active inference can provide a novel framework for modeling road user interactions which is also applicable in other fields.

Petr Vanc, Giovanni Franzese, Jan Kristof Behrens et al.

Learning from demonstration is a promising approach for teaching robots new skills. However, a central challenge in the execution of acquired skills is the ability to recognize faults and prevent failures. This is essential because demonstrations typically cover only a limited set of scenarios and often only the successful ones. During task execution, unforeseen situations may arise, such as changes in the robot's environment or interaction with human operators. To recognize such situations, this paper focuses on teaching the robot situational awareness by using a camera input and labeling frames as safe or risky. We train a Gaussian Process (GP) regression model fed by a low-dimensional latent space representation of the input images. The model outputs a continuous risk score ranging from zero to one, quantifying the degree of risk at each timestep. This allows for pausing task execution in unsafe situations and directly adding new training data, labeled by the human user. Our experiments on a robotic manipulator show that the proposed method can reliably detect both known and novel faults using only a single example for each new fault. In contrast, a standard multi-layer perceptron (MLP) performs well only on faults it has encountered during training. Our method enables the next generation of cobots to be rapidly deployed with easy-to-set-up, vision-based risk assessment, proactively safeguarding humans and detecting misaligned parts or missing objects before failures occur. We provide all the code and data required to reproduce our experiments at imitrob.ciirc.cvut.cz/publications/ilesia.

Rodrigo Pérez-Dattari, Zhaoting Li, Robert Babuška et al.

Planning methods struggle with computational intractability in solving task-level problems in large-scale environments. This work explores leveraging the commonsense knowledge encoded in LLMs to empower planning techniques to deal with these complex scenarios. We achieve this by efficiently using LLMs to prune irrelevant components from the planning problem's state space, substantially simplifying its complexity. We demonstrate the efficacy of this system through extensive experiments within a household simulation environment, alongside real-world validation using a 7-DoF manipulator (video https://youtu.be/6ro2UOtOQS4).

Jelle Luijkx, Runyu Ma, Zlatan Ajanović et al.

Reinforcement learning (RL) is a promising approach for robotic manipulation, but it can suffer from low sample efficiency and requires extensive exploration of large state-action spaces. Recent methods leverage the commonsense knowledge and reasoning abilities of large language models (LLMs) to guide exploration toward more meaningful states. However, LLMs can produce plans that are semantically plausible yet physically infeasible, yielding unreliable behavior. We introduce LLM-TALE, a framework that uses LLMs' planning to directly steer RL exploration. LLM-TALE integrates planning at both the task level and the affordance level, improving learning efficiency by directing agents toward semantically meaningful actions. Unlike prior approaches that assume optimal LLM-generated plans or rewards, LLM-TALE corrects suboptimality online and explores multimodal affordance-level plans without human supervision. We evaluate LLM-TALE on pick-and-place tasks in standard RL benchmarks, observing improvements in both sample efficiency and success rates over strong baselines. Real-robot experiments indicate promising zero-shot sim-to-real transfer. Code and supplementary material are available at llm-tale.github.io.

Julian F. Schumann, Eduardo Figueiredo, Frederik Baymler Mathiesen et al.

Accurate and robust trajectory prediction is essential for safe and efficient autonomous driving, yet recent work has shown that even state-of-the-art prediction models are highly vulnerable to inputs being mildly perturbed by adversarial attacks. Although model vulnerabilities to such attacks have been studied, work on effective countermeasures remains limited. In this work, we develop and evaluate a new defense mechanism for trajectory prediction models based on randomized smoothing -- an approach previously applied successfully in other domains. We evaluate its ability to improve model robustness through a series of experiments that test different strategies of randomized smoothing. We show that our approach can consistently improve prediction robustness of multiple base trajectory prediction models in various datasets without compromising accuracy in non-adversarial settings. Our results demonstrate that randomized smoothing offers a simple and computationally inexpensive technique for mitigating adversarial attacks in trajectory prediction.

Daniel Jarne Ornia, Giannis Delimpaltadakis, Jens Kober et al.

In Reinforcement Learning (RL), agents have no incentive to exhibit predictable behaviors, and are often pushed (through e.g. policy entropy regularisation) to randomise their actions in favor of exploration. This often makes it challenging for other agents and humans to predict an agent's behavior, triggering unsafe scenarios (e.g. in human-robot interaction). We propose a novel method to induce predictable behavior in RL agents, termed Predictability-Aware RL (PARL), employing the agent's trajectory entropy rate to quantify predictability. Our method maximizes a linear combination of a standard discounted reward and the negative entropy rate, thus trading off optimality with predictability. We show how the entropy rate can be formally cast as an average reward, how entropy-rate value functions can be estimated from a learned model and incorporate this in policy-gradient algorithms, and demonstrate how this approach produces predictable (near-optimal) policies in tasks inspired by human-robot use-cases.

Osama Mazhar, Jens Kober

In this paper, we propose a new data augmentation method, Random Shadows and Highlights (RSH) to acquire robustness against lighting perturbations. Our method creates random shadows and highlights on images, thus challenging the neural network during the learning process such that it acquires immunity against such input corruptions in real world applications. It is a parameter-learning free method which can be integrated into most vision related learning applications effortlessly. With extensive experimentation, we demonstrate that RSH not only increases the robustness of the models against lighting perturbations, but also reduces over-fitting significantly. Thus RSH should be considered essential for all vision related learning systems. Code is available at: https://github.com/OsamaMazhar/Random-Shadows-Highlights.

Antonin Raffin, Jens Kober, Freek Stulp

Reinforcement learning (RL) enables robots to learn skills from interactions with the real world. In practice, the unstructured step-based exploration used in Deep RL -- often very successful in simulation -- leads to jerky motion patterns on real robots. Consequences of the resulting shaky behavior are poor exploration, or even damage to the robot. We address these issues by adapting state-dependent exploration (SDE) to current Deep RL algorithms. To enable this adaptation, we propose two extensions to the original SDE, using more general features and re-sampling the noise periodically, which leads to a new exploration method generalized state-dependent exploration (gSDE). We evaluate gSDE both in simulation, on PyBullet continuous control tasks, and directly on three different real robots: a tendon-driven elastic robot, a quadruped and an RC car. The noise sampling interval of gSDE permits to have a compromise between performance and smoothness, which allows training directly on the real robots without loss of performance. The code is available at https://github.com/DLR-RM/stable-baselines3.

Yongxi Cao, Julian F. Schumann, Jens Kober et al.

Deep generative models such as conditional variational autoencoders (CVAEs) have shown great promise for predicting trajectories of surrounding agents in autonomous vehicle planning. State-of-the-art models have achieved remarkable accuracy in such prediction tasks. Besides accuracy, diversity is also crucial for safe planning because human behaviors are inherently uncertain and multimodal. However, existing methods generally lack a scheme to generate controllably diverse trajectories, which is arguably more useful than randomly diversified trajectories, to the end of safe planning. To address this, we propose PrefCVAE, an augmented CVAE framework that uses weakly labeled preference pairs to imbue latent variables with semantic attributes. Using average velocity as an example attribute, we demonstrate that PrefCVAE enables controllable, semantically meaningful predictions without degrading baseline accuracy. Our results show the effectiveness of preference supervision as a cost-effective way to enhance sampling-based generative models.

Julian F. Schumann, Anna Mészáros, Jens Kober et al.

While trajectory prediction plays a critical role in enabling safe and effective path-planning in automated vehicles, standardized practices for evaluating such models remain underdeveloped. Recent efforts have aimed to unify dataset formats and model interfaces for easier comparisons, yet existing frameworks often fall short in supporting heterogeneous traffic scenarios, joint prediction models, or user documentation. In this work, we introduce STEP -- a new benchmarking framework that addresses these limitations by providing a unified interface for multiple datasets, enforcing consistent training and evaluation conditions, and supporting a wide range of prediction models. We demonstrate the capabilities of STEP in a number of experiments which reveal 1) the limitations of widely-used testing procedures, 2) the importance of joint modeling of agents for better predictions of interactions, and 3) the vulnerability of current state-of-the-art models against both distribution shifts and targeted attacks by adversarial agents. With STEP, we aim to shift the focus from the ``leaderboard'' approach to deeper insights about model behavior and generalization in complex multi-agent settings.

Jelle Luijkx, Zlatan Ajanović, Laura Ferranti et al.

Human teaching effort is a significant bottleneck for the broader applicability of interactive imitation learning. To reduce the number of required queries, existing methods employ active learning to query the human teacher only in uncertain, risky, or novel situations. However, during these queries, the novice's planned actions are not utilized despite containing valuable information, such as the novice's capabilities, as well as corresponding uncertainty levels. To this end, we allow the novice to say: "I plan to do this, but I am uncertain." We introduce the Active Skill-level Data Aggregation (ASkDAgger) framework, which leverages teacher feedback on the novice plan in three key ways: (1) S-Aware Gating (SAG): Adjusts the gating threshold to track sensitivity, specificity, or a minimum success rate; (2) Foresight Interactive Experience Replay (FIER), which recasts valid and relabeled novice action plans into demonstrations; and (3) Prioritized Interactive Experience Replay (PIER), which prioritizes replay based on uncertainty, novice success, and demonstration age. Together, these components balance query frequency with failure incidence, reduce the number of required demonstration annotations, improve generalization, and speed up adaptation to changing domains. We validate the effectiveness of ASkDAgger through language-conditioned manipulation tasks in both simulation and real-world environments. Code, data, and videos are available at https://askdagger.github.io.

Maarten Hugenholtz, Anna Meszaros, Jens Kober et al.

Autonomous Vehicle decisions rely on multimodal prediction models that account for multiple route options and the inherent uncertainty in human behavior. However, models can suffer from mode collapse, where only the most likely mode is predicted, posing significant safety risks. While existing methods employ various strategies to generate diverse predictions, they often overlook the diversity in interaction modes among agents. Additionally, traditional metrics for evaluating prediction models are dataset-dependent and do not evaluate inter-agent interactions quantitatively. To our knowledge, none of the existing metrics explicitly evaluates mode collapse. In this paper, we propose a novel evaluation framework that assesses mode collapse in joint trajectory predictions, focusing on safety-critical interactions. We introduce metrics for mode collapse, mode correctness, and coverage, emphasizing the sequential dimension of predictions. By testing four multi-agent trajectory prediction models, we demonstrate that mode collapse indeed happens. When looking at the sequential dimension, although prediction accuracy improves closer to interaction events, there are still cases where the models are unable to predict the correct interaction mode, even just before the interaction mode becomes inevitable. We hope that our framework can help researchers gain new insights and advance the development of more consistent and accurate prediction models, thus enhancing the safety of autonomous driving systems.

Anish Abhijit Diwan, Julen Urain, Jens Kober et al.

This paper introduces a new imitation learning framework based on energy-based generative models capable of learning complex, physics-dependent, robot motion policies through state-only expert motion trajectories. Our algorithm, called Noise-conditioned Energy-based Annealed Rewards (NEAR), constructs several perturbed versions of the expert's motion data distribution and learns smooth, and well-defined representations of the data distribution's energy function using denoising score matching. We propose to use these learnt energy functions as reward functions to learn imitation policies via reinforcement learning. We also present a strategy to gradually switch between the learnt energy functions, ensuring that the learnt rewards are always well-defined in the manifold of policy-generated samples. We evaluate our algorithm on complex humanoid tasks such as locomotion and martial arts and compare it with state-only adversarial imitation learning algorithms like Adversarial Motion Priors (AMP). Our framework sidesteps the optimisation challenges of adversarial imitation learning techniques and produces results comparable to AMP in several quantitative metrics across multiple imitation settings.

Anna Mészáros, Julian F. Schumann, Javier Alonso-Mora et al.

The estimation of probability density functions is a fundamental problem in science and engineering. However, common methods such as kernel density estimation (KDE) have been demonstrated to lack robustness, while more complex methods have not been evaluated in multi-modal estimation problems. In this paper, we present ROME (RObust Multi-modal Estimator), a non-parametric approach for density estimation which addresses the challenge of estimating multi-modal, non-normal, and highly correlated distributions. ROME utilizes clustering to segment a multi-modal set of samples into multiple uni-modal ones and then combines simple KDE estimates obtained for individual clusters in a single multi-modal estimate. We compared our approach to state-of-the-art methods for density estimation as well as ablations of ROME, showing that it not only outperforms established methods but is also more robust to a variety of distributions. Our results demonstrate that ROME can overcome the issues of over-fitting and over-smoothing exhibited by other estimators.

Julian F. Schumann, Aravinda Ramakrishnan Srinivasan, Jens Kober et al.

The development of automated vehicles has the potential to revolutionize transportation, but they are currently unable to ensure a safe and time-efficient driving style. Reliable models predicting human behavior are essential for overcoming this issue. While data-driven models are commonly used to this end, they can be vulnerable in safety-critical edge cases. This has led to an interest in models incorporating cognitive theory, but as such models are commonly developed for explanatory purposes, this approach's effectiveness in behavior prediction has remained largely untested so far. In this article, we investigate the usefulness of the \emph{Commotions} model -- a novel cognitively plausible model incorporating the latest theories of human perception, decision-making, and motor control -- for predicting human behavior in gap acceptance scenarios, which entail many important traffic interactions such as lane changes and intersections. We show that this model can compete with or even outperform well-established data-driven prediction models across several naturalistic datasets. These results demonstrate the promise of incorporating cognitive theory in behavior prediction models for automated vehicles.

Jihong Zhu, Michael Gienger, Jens Kober

Moving away from repetitive tasks, robots nowadays demand versatile skills that adapt to different situations. Task-parameterized learning improves the generalization of motion policies by encoding relevant contextual information in the task parameters, hence enabling flexible task executions. However, training such a policy often requires collecting multiple demonstrations in different situations. To comprehensively create different situations is non-trivial thus renders the method less applicable to real-world problems. Therefore, training with fewer demonstrations/situations is desirable. This paper presents a novel concept to augment the original training dataset with synthetic data for policy improvements, thus allows learning task-parameterized skills with few demonstrations.

Anna Mészáros, Giovanni Franzese, Jens Kober

This work investigates how the intricate task of a continuous pick & place (P&P) motion may be learned from humans based on demonstrations and corrections. Due to the complexity of the task, these demonstrations are often slow and even slightly flawed, particularly at moments when multiple aspects (i.e., end-effector movement, orientation, and gripper width) have to be demonstrated at once. Rather than training a person to give better demonstrations, non-expert users are provided with the ability to interactively modify the dynamics of their initial demonstration through teleoperated corrective feedback. This in turn allows them to teach motions outside of their own physical capabilities. In the end, the goal is to obtain a faster but reliable execution of the task. The presented framework learns the desired movement dynamics based on the current Cartesian position with Gaussian Processes (GPs), resulting in a reactive, time-invariant policy. Using GPs also allows online interactive corrections and active disturbance rejection through epistemic uncertainty minimization. The experimental evaluation of the framework is carried out on a Franka-Emika Panda. Tests were performed to determine i) the framework's effectiveness in successfully learning how to quickly pick & place an object, ii) ease of policy correction to environmental changes (i.e., different object sizes and mass), and iii) the framework's usability for non-expert users.

Jihong Zhu, Andrea Cherubini, Claire Dune et al.

Deformable object manipulation (DOM) is an emerging research problem in robotics. The ability to manipulate deformable objects endows robots with higher autonomy and promises new applications in the industrial, services, and healthcare sectors. However, compared to rigid object manipulation, the manipulation of deformable objects is considerably more complex, and is still an open research problem. Addressing DOM challenges demand breakthroughs in almost all aspects of robotics, namely hardware design, sensing, (deformation) modeling, planning, and control. In this article, we review recent advances and highlight the main challenges when considering deformation in each sub-field. A particular focus of our paper lies in the discussions of these challenges and proposing future directions of research.

Giovanni Franzese, Anna Mészáros, Luka Peternel et al.

Teaching robots how to apply forces according to our preferences is still an open challenge that has to be tackled from multiple engineering perspectives. This paper studies how to learn variable impedance policies where both the Cartesian stiffness and the attractor can be learned from human demonstrations and corrections with a user-friendly interface. The presented framework, named ILoSA, uses Gaussian Processes for policy learning, identifying regions of uncertainty and allowing interactive corrections, stiffness modulation and active disturbance rejection. The experimental evaluation of the framework is carried out on a Franka-Emika Panda in four separate cases with unique force interaction properties: 1) pulling a plug wherein a sudden force discontinuity occurs upon successful removal of the plug, 2) pushing a box where a sustained force is required to keep the robot in motion, 3) wiping a whiteboard in which the force is applied perpendicular to the direction of movement, and 4) inserting a plug to verify the usability for precision-critical tasks in an experimental validation performed with non-expert users.

Osama Mazhar, Robert Babuska, Jens Kober

Deep neural networks designed for vision tasks are often prone to failure when they encounter environmental conditions not covered by the training data. Single-modal strategies are insufficient when the sensor fails to acquire information due to malfunction or its design limitations. Multi-sensor configurations are known to provide redundancy, increase reliability, and are crucial in achieving robustness against asymmetric sensor failures. To address the issue of changing lighting conditions and asymmetric sensor degradation in object detection, we develop a multi-modal 2D object detector, and propose deterministic and stochastic sensor-aware feature fusion strategies. The proposed fusion mechanisms are driven by the estimated sensor measurement reliability values/weights. Reliable object detection in harsh lighting conditions is essential for applications such as self-driving vehicles and human-robot interaction. We also propose a new "r-blended" hybrid depth modality for RGB-D sensors. Through extensive experimentation, we show that the proposed strategies outperform the existing state-of-the-art methods on the FLIR-Thermal dataset, and obtain promising results on the SUNRGB-D dataset. We additionally record a new RGB-Infra indoor dataset, namely L515-Indoors, and demonstrate that the proposed object detection methodologies are highly effective for a variety of lighting conditions.

Bas van der Heijden, Laura Ferranti, Jens Kober et al.

This paper presents DeepKoCo, a novel model-based agent that learns a latent Koopman representation from images. This representation allows DeepKoCo to plan efficiently using linear control methods, such as linear model predictive control. Compared to traditional agents, DeepKoCo learns task-relevant dynamics, thanks to the use of a tailored lossy autoencoder network that allows DeepKoCo to learn latent dynamics that reconstruct and predict only observed costs, rather than all observed dynamics. As our results show, DeepKoCo achieves similar final performance as traditional model-free methods on complex control tasks while being considerably more robust to distractor dynamics, making the proposed agent more amenable for real-life applications.

Snehal Jauhri, Carlos Celemin, Jens Kober

Imitation Learning techniques enable programming the behavior of agents through demonstrations rather than manual engineering. However, they are limited by the quality of available demonstration data. Interactive Imitation Learning techniques can improve the efficacy of learning since they involve teachers providing feedback while the agent executes its task. In this work, we propose a novel Interactive Learning technique that uses human feedback in state-space to train and improve agent behavior (as opposed to alternative methods that use feedback in action-space). Our method titled Teaching Imitative Policies in State-space~(TIPS) enables providing guidance to the agent in terms of `changing its state' which is often more intuitive for a human demonstrator. Through continuous improvement via corrective feedback, agents trained by non-expert demonstrators using TIPS outperformed the demonstrator and conventional Imitation Learning agents.

Rodrigo Pérez-Dattari, Carlos Celemin, Javier Ruiz-del-Solar et al.

Deep Reinforcement Learning (DRL) has become a powerful methodology to solve complex decision-making problems. However, DRL has several limitations when used in real-world problems (e.g., robotics applications). For instance, long training times are required and cannot be accelerated in contrast to simulated environments, and reward functions may be hard to specify/model and/or to compute. Moreover, the transfer of policies learned in a simulator to the real-world has limitations (reality gap). On the other hand, machine learning methods that rely on the transfer of human knowledge to an agent have shown to be time efficient for obtaining well performing policies and do not require a reward function. In this context, we analyze the use of human corrective feedback during task execution to learn policies with high-dimensional state spaces, by using the D-COACH framework, and we propose new variants of this framework. D-COACH is a Deep Learning based extension of COACH (COrrective Advice Communicated by Humans), where humans are able to shape policies through corrective advice. The enhanced version of D-COACH, which is proposed in this paper, largely reduces the time and effort of a human for training a policy. Experimental results validate the efficiency of the D-COACH framework in three different problems (simulated and with real robots), and show that its enhanced version reduces the human training effort considerably, and makes it feasible to learn policies within periods of time in which a DRL agent do not reach any improvement.

Jan Scholten, Daan Wout, Carlos Celemin et al.

Deep Reinforcement Learning has enabled the control of increasingly complex and high-dimensional problems. However, the need of vast amounts of data before reasonable performance is attained prevents its widespread application. We employ binary corrective feedback as a general and intuitive manner to incorporate human intuition and domain knowledge in model-free machine learning. The uncertainty in the policy and the corrective feedback is combined directly in the action space as probabilistic conditional exploration. As a result, the greatest part of the otherwise ignorant learning process can be avoided. We demonstrate the proposed method, Predictive Probabilistic Merging of Policies (PPMP), in combination with DDPG. In experiments on continuous control problems of the OpenAI Gym, we achieve drastic improvements in sample efficiency, final performance, and robustness to erroneous feedback, both for human and synthetic feedback. Additionally, we show solutions beyond the demonstrated knowledge.

Daan Wout, Jan Scholten, Carlos Celemin et al.

Learning from human feedback is a viable alternative to control design that does not require modelling or control expertise. Particularly, learning from corrective advice garners advantages over evaluative feedback as it is a more intuitive and scalable format. The current state-of-the-art in this field, COACH, has proven to be a effective approach for confined problems. However, it parameterizes the policy with Radial Basis Function networks, which require meticulous feature space engineering for higher order systems. We introduce Gaussian Process Coach (GPC), where feature space engineering is avoided by employing Gaussian Processes. In addition, we use the available policy uncertainty to 1) inquire feedback samples of maximal utility and 2) to adapt the learning rate to the teacher's learning phase. We demonstrate that the novel algorithm outperforms the current state-of-the-art in final performance, convergence rate and robustness to erroneous feedback in OpenAI Gym continuous control benchmarks, both for simulated and real human teachers.

Rodrigo Pérez-Dattari, Carlos Celemin, Javier Ruiz-del-Solar et al.

Deep Reinforcement Learning (DRL) has become a powerful strategy to solve complex decision making problems based on Deep Neural Networks (DNNs). However, it is highly data demanding, so unfeasible in physical systems for most applications. In this work, we approach an alternative Interactive Machine Learning (IML) strategy for training DNN policies based on human corrective feedback, with a method called Deep COACH (D-COACH). This approach not only takes advantage of the knowledge and insights of human teachers as well as the power of DNNs, but also has no need of a reward function (which sometimes implies the need of external perception for computing rewards). We combine Deep Learning with the COrrective Advice Communicated by Humans (COACH) framework, in which non-expert humans shape policies by correcting the agent's actions during execution. The D-COACH framework has the potential to solve complex problems without much data or time required. Experimental results validated the efficiency of the framework in three different problems (two simulated, one with a real robot), with state spaces of low and high dimensions, showing the capacity to successfully learn policies for continuous action spaces like in the Car Racing and Cart-Pole problems faster than with DRL.

Jens Kober, Jan Peters

Many tasks in robotics can be decomposed into sub-tasks that are performed simultaneously. In many cases, these sub-tasks cannot all be achieved jointly and a prioritization of such sub-tasks is required to resolve this issue. In this paper, we discuss a novel learning approach that allows to learn a prioritized control law built on a set of sub-tasks represented by motor primitives. The primitives are executed simultaneously but have different priorities. Primitives of higher priority can override the commands of the conflicting lower priority ones. The dominance structure of these primitives has a significant impact on the performance of the prioritized control law. We evaluate the proposed approach with a ball bouncing task on a Barrett WAM.