Martina Truschzinski

An Emotional Framework for a Real-Life Worker Simulation

Within the framework of the project ‘The Smart Virtual Worker’ we put forward a sound and functioning emotional model which adequately simulates a worker’s emotional feelings throughout a typical task in an industrial setting. We restricted the model to represent the basic emotions by Ekman and focused on the implementation of ‘joy’ and ‘anger’. Since emotions are uniquely generated, based on the interpretation of a stimulus by an individual, we linked the genesis of emotions to empirical findings of the sports sciences to infer an emotional reaction. This paper describes the concept of the model from a theoretical and practical point of view as well as the preliminary state of implementation and upcoming steps of the project.

Deducing human emotions by robots: Computing basic non-verbal expressions of performed actions during a work task

We have established an emotional model to enhance a virtual worker simulation, which could be also used to support robots in a joined human-robot work-task inside an industrial setting. The robot is able to understand peoples individual and specific knowledge as well as capabilities, which are ultimately linked to an emotional consequence. As a result, the emotional model outputs the emotional valence calculated as positive or negative values, respective to reward and punishment. This output is applied as value function for a reinforcement learning agent. There we use an actor critic algorithm extended by eligibility traces and task specific conditions to learn the optimal action sequences. We show the influence of emotional reward leads to differences in the learned action sequences in comparison to a simple task performance evaluation reward. Therefore the robot is able to calculate emotional feelings of a human during a given working task, is able to decide if there is a better, more emotional stable path to doing this working task and moreover the robot is able to decide when the human is needed help or even not.

Investigating the Influence of Emotion in Air Traffic Controller Tasks: Pretest Evaluation

Air traffic controllers (ATC) have to perform under highly demanding circumstances, and they are often responsible for the lives of more than 5.000 people in mid-air at any given moment. Therefore, besides regular scheduled breaks to recover, it is of paramount importance to be in a level-minded state in order to maximize cognitive capabilities. Within this context and based on our previous research with active air traffic controllers, we designed a comparable scenario in which two airplanes from different directions converge and where subjects have to prevent collision. They watched an emotion inducing video shortly before participating in the experiment. The induced positive or negative valence lead to significant group-differences during their performance of the ATC task.

Modeling Workload: A System Theory Approach

We present preliminary results in modeling workload with a dynamic process model. We build and validated our dynamical workload model on data of real pupil dilatation recorded during an conducted air traffic controller task experiment.

The Smart Virtual Worker – Digitales Menschmodell für die Simulation industrieller Arbeitsvorgänge

Im Rahmen eines interdisziplinaren Forschungsprojekts an der Technischen Universitat Chemnitz wurde das digitale Menschmodell „The Smart Virtual Worker“ (SVW) entwickelt. Mit der Entwicklung werden eine signifikante Reduktion im Simulationsaufwand sowie eine Erweiterung integrierter arbeitswissenschaftlicher Bewertungsmethoden in digitalen Menschmodellen angestrebt. Der Beitrag erklart die dazu im SVW integrierten Module und Methoden sowie deren Zusammenwirken. Im SVW erfolgt die Erstellung des Arbeitsprozesses mittels autonomer Handlungsselektion unter Nutzung von Kunstlichen Intelligenz (KI) – Algorithmen in Kombination mit einer automatischen Bewegungserzeugung. Der Nutzer kann hier entscheiden, ob der Arbeitsprozess nach zeitlichen, ergonomischen oder emotionsbasierten Kriterien optimiert wird. Die ausgewahlten Handlungen werden anschliesend in Bewegungsablaufe ubersetzt. Zur Bewertung des erzeugten Arbeitsprozesses stehen Verfahren zur ergonomischen Bewertung der Arbeitsaufgabe (RULA-/EAWS-Verfahren) und Arbeitsumwelt (Larm-, Klima- und Beleuchtungsbewertung) sowie zur psychologischen Bewertung (Emotionsmodell) zur Verfugung. Im Ergebnis konnte der Erstellungsaufwand fur eine Arbeitsprozesssimulation wesentlich reduziert und eine Erweiterung der arbeitswissenschaftlichen Bewertungsmethoden in DMM vorgenommen werden.

Analytical Steps for the Calibration of an Emotional Framework

The emotion model of the Smart Virtual Worker is the result of three years of interdisciplinary research. After successful implementation and pre-validation of the model and the surrounding simulation architecture, the model had to be calibrated by using real life working scenarios. The task of carrying differently weighed boxes over a 30 m distance was chosen as the foundation for the model. Subsequent fitting of the model led to a positive evaluation outcome which presented a mean 88 % fitting of the model’s simulated emotional valence in relation to the observed real world behavior.

An emotional model for social robots: late-breaking report

We developed an emotional model, which could help supporting robots to accomodate humans during a working task inside an industrial setting. The robot would recognize when a human is experiencing increased stress and decides whether it should assist the human or should do other tasks. We propose the model as a framework which was developed as part of“The Smart Virtual Worker“-project within the context of human-robot interactions. The emotional model is able to estimate a workers emotional valence throughout a typical work task by applying a hierarchical reinforcement learning algorithm. Since emotions are generated by the human brain based on an individuals interpretation of a stimulus, we linked the genesis of emotions to empirical findings of the sports sciences in order to infer an emotional reaction. Furthermore, the model reproduces sympathetic reactions of the human body and is capable of remembering past actions in order to include possible future time constraints as an initiator for emotional responses in the upcoming iterations. This capability is crucial for accommodating long-term experiences since the emotional reaction is not only based on the present situation, but on the whole experimental setting.