Karla Stepanova

Petr Vanc, Jan Kristof Behrens, Václav Hlaváč et al.

Instructing robots from demonstrations can be done through different teaching modalities, each with different usability and performance trade-offs. This paper compares kinesthetic guidance, joystick teleoperation, and hand gestures in a user study with eight participants. We evaluate replay success, modified NASA-TLX workload, and common teaching errors across three manipulation tasks. Kinesthetic guidance produced the shortest demonstrations, lowest workload, and highest success on the more orientation-sensitive and contact-rich tasks. Joystick teleoperation performed best on simple peg picking. Hand-gesture teaching, although less reliable overall, performed better than expected and in some cases achieved results comparable to kinesthetic guidance.

Jiri Sedlar, Karla Stepanova, Radoslav Skoviera et al.

This paper introduces a dataset for training and evaluating methods for 6D pose estimation of hand-held tools in task demonstrations captured by a standard RGB camera. Despite the significant progress of 6D pose estimation methods, their performance is usually limited for heavily occluded objects, which is a common case in imitation learning, where the object is typically partially occluded by the manipulating hand. Currently, there is a lack of datasets that would enable the development of robust 6D pose estimation methods for these conditions. To overcome this problem, we collect a new dataset (Imitrob) aimed at 6D pose estimation in imitation learning and other applications where a human holds a tool and performs a task. The dataset contains image sequences of nine different tools and twelve manipulation tasks with two camera viewpoints, four human subjects, and left/right hand. Each image is accompanied by an accurate ground truth measurement of the 6D object pose obtained by the HTC Vive motion tracking device. The use of the dataset is demonstrated by training and evaluating a recent 6D object pose estimation method (DOPE) in various setups.

Oleh Borys, Karla Stepanova

Learning from Demonstration~(LfD) should capture not only how a task is executed, but also its high-level task structure that explains the demonstrated behavior. As robots become more autonomous, such task representations must be inspectable, reusable, and human-interpretable. To address this, we study how to represent and learn robotic tasks with inductive logic programming~(ILP) by decomposing a complex task into a series of simpler learning objectives at different abstraction (ontological) levels. The system infers symbolic rules from demonstrations and prior (domain) knowledge, and reuses learned rules when learning higher-level task structure. We evaluate the approach in a synthetic block-assembly scenario and show that the learned abstractions are interpretable and support strong generalization to harder, held-out tasks with unseen objects. These results provide preliminary evidence that decomposed ILP is a feasible approach to task-level LfD.

Miroslav David, Karla Stepanova, Robert Babuska

Robotic surface-interaction tasks, such as spray painting or welding, require both accurate geometric planning and precise motion execution. While modern motion planners generate valid geometric paths, they often lack the expert motor patterns observed in human operators. Conversely, learning from demonstration often tightly couples task execution to the specific training geometry, limiting transferability. We propose a modular framework that decouples geometric motion planning from execution-level expertise. Expert behavior is represented as a vocabulary of interpretable, atomic motor rules, such as velocity scaling and orientation offsets, that systematically modify a geometrically planned reference path. We train a multimodal neural network to infer rule parameters jointly from kinematic trajectory data and CAD model geometry. We evaluate our approach through dynamic simulation on L-shaped and window-shaped objects, demonstrating on simulated data that the model successfully extracts velocity and orientation rules across both topologies.

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.

Gabriela Sejnova, Michal Vavrecka, Karla Stepanova et al.

Multimodal Variational Autoencoders (VAEs) have been the subject of intense research in the past years as they can integrate multiple modalities into a joint representation and can thus serve as a promising tool for both data classification and generation. Several approaches toward multimodal VAE learning have been proposed so far, their comparison and evaluation have however been rather inconsistent. One reason is that the models differ at the implementation level, another problem is that the datasets commonly used in these cases were not initially designed to evaluate multimodal generative models. This paper addresses both mentioned issues. First, we propose a toolkit for systematic multimodal VAE training and comparison. The toolkit currently comprises 4 existing multimodal VAEs and 6 commonly used benchmark datasets along with instructions on how to easily add a new model or a dataset. Second, we present a disentangled bimodal dataset designed to comprehensively evaluate the joint generation and cross-generation capabilities across multiple difficulty levels. We demonstrate the utility of our dataset by comparing the implemented state-of-the-art models.

Michal Nazarczuk, Jan Kristof Behrens, Karla Stepanova et al.

Embodied reasoning systems integrate robotic hardware and cognitive processes to perform complex tasks, typically in response to a natural language query about a specific physical environment. This usually involves changing the belief about the scene or physically interacting and changing the scene (e.g. sort the objects from lightest to heaviest). In order to facilitate the development of such systems we introduce a new modular Closed Loop Interactive Embodied Reasoning (CLIER) approach that takes into account the measurements of non-visual object properties, changes in the scene caused by external disturbances as well as uncertain outcomes of robotic actions. CLIER performs multi-modal reasoning and action planning and generates a sequence of primitive actions that can be executed by a robot manipulator. Our method operates in a closed loop, responding to changes in the environment. Our approach is developed with the use of MuBle simulation environment and tested in 10 interactive benchmark scenarios. We extensively evaluate our reasoning approach in simulation and in real-world manipulation tasks with a success rate above 76% and 64%, respectively.

Gabriela Sejnova, Michal Vavrecka, Karla Stepanova

Variational Autoencoders (VAEs) are powerful generative models that have been widely used in various fields, including image and text generation. However, one of the known challenges in using VAEs is the model's sensitivity to its hyperparameters, such as the latent space size. This paper presents a simple extension of VAEs for automatically determining the optimal latent space size during the training process by gradually decreasing the latent size through neuron removal and observing the model performance. The proposed method is compared to traditional hyperparameter grid search and is shown to be significantly faster while still achieving the best optimal dimensionality on four image datasets. Furthermore, we show that the final performance of our method is comparable to training on the optimal latent size from scratch, and might thus serve as a convenient substitute.

Gabriela Sejnova, Michal Vavrecka, Karla Stepanova

In this work, we focus on unsupervised vision-language-action mapping in the area of robotic manipulation. Recently, multiple approaches employing pre-trained large language and vision models have been proposed for this task. However, they are computationally demanding and require careful fine-tuning of the produced outputs. A more lightweight alternative would be the implementation of multimodal Variational Autoencoders (VAEs) which can extract the latent features of the data and integrate them into a joint representation, as has been demonstrated mostly on image-image or image-text data for the state-of-the-art models. Here we explore whether and how can multimodal VAEs be employed in unsupervised robotic manipulation tasks in a simulated environment. Based on the obtained results, we propose a model-invariant training alternative that improves the models' performance in a simulator by up to 55%. Moreover, we systematically evaluate the challenges raised by the individual tasks such as object or robot position variability, number of distractors or the task length. Our work thus also sheds light on the potential benefits and limitations of using the current multimodal VAEs for unsupervised learning of robotic motion trajectories based on vision and language.

Petr Vanc, Jan Kristof Behrens, Václav Hlaváč et al.

Programming robots by demonstration (PbD) is an intuitive concept, but scaling it to real-world variability remains a challenge for most current teaching frameworks. Conditional task graphs are very expressive and can be defined incrementally, which fits very well with the PbD idea. However, acting using conditional task graphs requires reliable perception-grounded online branch selection. In this paper, we present See & Switch, an interactive teaching-and-execution framework that represents tasks as user-extendable graphs of skill parts connected via decision states (DS), enabling conditional branching during replay. Unlike prior approaches that rely on manual branching or low-dimensional signals (e.g., proprioception), our vision-based Switcher uses eye-in-hand images (high-dimensional) to select among competing successor skill parts and to detect out-of-distribution contexts that require new demonstrations. We integrate kinesthetic teaching, joystick control, and hand gestures via an input-modality-abstraction layer and demonstrate that our proposed method is teaching modality-independent, enabling efficient in-situ recovery demonstrations. The system is validated in experiments on three challenging dexterous manipulation tasks. We evaluate our method under diverse conditions and furthermore conduct user studies with 8 participants. We show that the proposed method reliably performs branch selection and anomaly detection for novice users, achieving 90.7 % and 87.9 % accuracy, respectively, across 576 real-robot rollouts. We provide all code and data required to reproduce our experiments at http://imitrob.ciirc.cvut.cz/publications/seeandswitch.

Petr Vanc, Karla Stepanova

As human-robot collaboration advances, natural and flexible communication methods are essential for effective robot control. Traditional methods relying on a single modality or rigid rules struggle with noisy or misaligned data as well as with object descriptions that do not perfectly fit the predefined object names (e.g. 'Pick that red object'). We introduce TransforMerger, a transformer-based reasoning model that infers a structured action command for robotic manipulation based on fused voice and gesture inputs. Our approach merges multimodal data into a single unified sentence, which is then processed by the language model. We employ probabilistic embeddings to handle uncertainty and we integrate contextual scene understanding to resolve ambiguous references (e.g., gestures pointing to multiple objects or vague verbal cues like "this"). We evaluate TransforMerger in simulated and real-world experiments, demonstrating its robustness to noise, misalignment, and missing information. Our results show that TransforMerger outperforms deterministic baselines, especially in scenarios requiring more contextual knowledge, enabling more robust and flexible human-robot communication. Code and datasets are available at: http://imitrob.ciirc.cvut.cz/publications/transformerger.

Karla Stepanova, Jakub Rozlivek, Frantisek Puciow et al.

We present a robot kinematic calibration method that combines complementary calibration approaches: self-contact, planar constraints, and self-observation. We analyze the estimation of the end effector parameters, joint offsets of the manipulators, and calibration of the complete kinematic chain (DH parameters). The results are compared with ground truth measurements provided by a laser tracker. Our main findings are: (1) When applying the complementary calibration approaches in isolation, the self-contact approach yields the best and most stable results. (2) All combinations of more than one approach were always superior to using any single approach in terms of calibration errors and the observability of the estimated parameters. Combining more approaches delivers robot parameters that better generalize to the workspace parts not used for the calibration. (3) Sequential calibration, i.e. calibrating cameras first and then robot kinematics, is more effective than simultaneous calibration of all parameters. In real experiments, we employ two industrial manipulators mounted on a common base. The manipulators are equipped with force/torque sensors at their wrists, with two cameras attached to the robot base, and with special end effectors with fiducial markers. We collect a new comprehensive dataset for robot kinematic calibration and make it publicly available. The dataset and its analysis provide quantitative and qualitative insights that go beyond the specific manipulators used in this work and apply to self-contained robot kinematic calibration in general.

Radoslav Skoviera, Karla Stepanova, Michael Tesar et al.

In this paper, we consider the problem of learning object manipulation tasks from human demonstration using RGB or RGB-D cameras. We highlight the key challenges in capturing sufficiently good data with no tracking devices - starting from sensor selection and accurate 6DoF pose estimation to natural language processing. In particular, we focus on two showcases: gluing task with a glue gun and simple block-stacking with variable blocks. Furthermore, we discuss how a linguistic description of the task could help to improve the accuracy of task description. We also present the whole architecture of our transfer of the imitated task to the simulated and real robot environment.

Karla Stepanova, Tomas Pajdla, Matej Hoffmann

Mechanism calibration is an important and non-trivial task in robotics. Advances in sensor technology make affordable but increasingly accurate devices such as cameras and tactile sensors available, making it possible to perform automated self-contained calibration relying on redundant information in these sensory streams. In this work, we use a simulated iCub humanoid robot with a stereo camera system and end-effector contact emulation to quantitatively compare the performance of kinematic calibration by employing different combinations of intersecting kinematic chains -- either through self-observation or self-touch. The parameters varied were: (i) type and number of intersecting kinematic chains used for calibration, (ii) parameters and chains subject to optimization, (iii) amount of initial perturbation of kinematic parameters, (iv) number of poses/configurations used for optimization, (v) amount of measurement noise in end-effector positions / cameras. The main findings are: (1) calibrating parameters of a single chain (e.g. one arm) by employing multiple kinematic chains ("self-observation" and "self-touch") is superior in terms of optimization results as well as observability; (2) when using multi-chain calibration, fewer poses suffice to get similar performance compared to when for example only observation from a single camera is used; (3) parameters of all chains (here 86 DH parameters) can be subject to calibration simultaneously and with 50 (100) poses, end-effector error of around 2 (1) mm can be achieved; (4) adding noise to a sensory modality degrades performance of all calibrations employing the chains relying on this information.

Karla Stepanova, Matej Hoffmann, Zdenek Straka et al.

Humans and animals are constantly exposed to a continuous stream of sensory information from different modalities. At the same time, they form more compressed representations like concepts or symbols. In species that use language, this process is further structured by this interaction, where a mapping between the sensorimotor concepts and linguistic elements needs to be established. There is evidence that children might be learning language by simply disambiguating potential meanings based on multiple exposures to utterances in different contexts (cross-situational learning). In existing models, the mapping between modalities is usually found in a single step by directly using frequencies of referent and meaning co-occurrences. In this paper, we present an extension of this one-step mapping and introduce a newly proposed sequential mapping algorithm together with a publicly available Matlab implementation. For demonstration, we have chosen a less typical scenario: instead of learning to associate objects with their names, we focus on body representations. A humanoid robot is receiving tactile stimulations on its body, while at the same time listening to utterances of the body part names (e.g., hand, forearm and torso). With the goal at arriving at the correct "body categories", we demonstrate how a sequential mapping algorithm outperforms one-step mapping. In addition, the effect of data set size and noise in the linguistic input are studied.