Tomasz Grzejszczak

This paper investigates human-robot interaction (HRI) within the challenging, high-noise environments of educational fairs, advocating for an HCI-centered approach to evaluate social robots in unpredictable public spaces. Utilizing the advanced social robot Furhat, the study compares the effectiveness of speech-based interaction against gesture-based recognition to overcome the limitations of traditional voice systems in loud settings. Five interaction programs were evaluated: a Large Language Model (LLM) variant relying on voice, and two gesture-controlled games (Rock-Paper-Scissors and Blocks guessing game) tested both with and without passive robot gestures. The results, drawn from field tests, demonstrate that the gesture recognition module is a highly effective alternative to speech recognition in noisy environments. While the voice-based LLM program struggled with a 33% success rate and high idle times due to environmental noise, gesture-based interactions achieved significantly higher success rates, ranging from 77% to 96%. Furthermore, the study confirms that a gesturing social robot is significantly more effective at attracting attention. The inclusion of passive gestures reduced the robot idle time from an average of 141.7–143.7 seconds to 105.6–106.5 seconds, while increasing participant engagement by 16% to 21%. These findings underscore the importance of non-verbal communication and multimodal perception in fostering reliable and engaging HRI in dynamic, high-social environments.

In recent years, the development of socially interactive robots has heightened the need for interpretable and trustworthy models capable of recognizing complex human emotions in real time. While deep convolutional neural networks achieve strong performance in facial affect recognition, their decision processes remain largely opaque, which limits user trust and safe deployment in human–robot interaction (HRI). This challenge becomes particularly critical when recognizing complex or subtle emotional states under domain shift, where both performance robustness and explanation stability are required. A robot-mediated explainability framework for complex emotion recognition links the visual attribut of a convolutional neural network (CNN) to social-robot behaviors in real time. The perception module augments a CNN with an Attention Map Alignment Layer (AMAL) that regularizes the internal focus toward semantically meaningful facial regions during training. Model-agnostic eXplainable AI (XAI) methods (SHAP and LIME) provide per-instance rationales at inference and are audited for temporal stability and faithfulness under streaming conditions. Explanations are embodied through a behavior-generation layer that converts saliency into deictic head motion and gaze, optionally governed by an adaptive emphasis policy that modulates when and how strongly to highlight evidence as a function of model confidence, explanation stability, and user-state signals. Evaluation spans AffectNet and EMOTIC, plus an out-of-distribution robot-domain set (OhBot) to probe domain shift. A human–robot interaction study uses a 3×2×2 between-subjects design crossing embodiment (no robot, still gaze, head motion), visual XAI (no heatmap vs heatmap), and emphasis policy (fixed vs adaptive). Across in-distribution and robot-domain tests, AMAL improves accuracy and macro-F1 with lower expected calibration error, and yields higher attribution–landmark overlap with greater temporal stability. In interaction, head motion increases acceptance of system recommendations relative to still-gaze and no-robot baselines, while heatmaps improve task performance irrespective of embodiment. Adaptive emphasis delivers additional gains primarily under well-calibrated predictions and attenuates guidance under low confidence, limiting over-reliance. Overall, aligning internal attention, surfacing faithful rationales, and embodying those rationales through socially legible behaviors enhances both objective performance and calibrated trust. Next steps include ondevice deployment on embedded accelerators, longitudinal evaluation of trust dynamics, and integration with language-model-based social-norm explanation modules as a complementary, normative layer.

The paper presents a simulation investigation into the individual time of parcel transportation in a multi-agent robot operated warehouse for two different artificial intelligence planning strategies (independent and cooperative), two different warehouse setups and different number of robots...

Controlling a remotely operated underwater vehicle (ROV) is an extremely challenging task that requires precise maneuvering and navigation in complex and often unpredictable environments. The operator faces numerous difficulties, including limited visibility and communication constraints, and the...

In this paper, surface electromyography (sEMG) is used to gather the activation neural signal from muscles during an indoor rowing exercise. The exercise was performed by professional athletes and amateur non-athletes. The data acquisition and processing are described to obtain a set of parameters: number of cycles, average cycle time, cycle time standard deviation, fatigue time, muscle activation time, and muscle energy. These parameters are used to draw conclusions on common non-athletes’ mistakes during exercise for better training advice and a way of statistically distinguishing an athlete from a non-athlete.

The field of research of this paper combines Human Computer Interface, gesture recognition and fingertips tracking. Most gesture recognition algorithms processing color images are unable to locate folded fingers hidden inside hand contour. With use of hand landmarks detection and localization algorithm, processing directional images, the fingertips are tracked whether they are risen or folded inside the hand contour. The capabilities of the method, repeatibility and accuracy, are tested with use of 3 gestures that are recorded on the USB camera. Fingertips are tracked in gestures presenting a linear movement of an open hand, finger folding into fist and clenched fist movement. In conclusion, a discussion of accuracy in application to HCI is presented.