Predicting Mid-Air Interaction Movements and Fatigue Using Deep Reinforcement Learning
| Autor: | Kourosh Naderi, Philipp Slusallek, Perttu Hämäläinen, Noshaba Cheema, Jaakko Lehtinen, Laura Frey-Law |
|---|---|
| Přispěvatelé: | Max Planck Institute for Informatics, University of Iowa, Professorship Hämäläinen Perttu, Professorship Lehtinen Jaakko, Saarland University, Department of Computer Science, Department of Media, Aalto-yliopisto, Aalto University |
| Rok vydání: | 2020 |
| Předmět: |
reinforcement learning
user modeling Computer science business.industry User modeling 05 social sciences Work (physics) 020207 software engineering Robotics 02 engineering and technology Task (computing) Control theory biomechanical simulation 0202 electrical engineering electronic engineering information engineering Torque Reinforcement learning 0501 psychology and cognitive sciences Artificial intelligence business computational interaction 050107 human factors Computer animation Simulation |
| Zdroj: | CHI |
| Popis: | A common problem of mid-air interaction is excessive arm fatigue, known as the "Gorilla arm" effect. To predict and prevent such problems at a low cost, we investigate user testing of mid-air interaction without real users, utilizing biomechanically simulated AI agents trained using deep Reinforcement Learning (RL). We implement this in a pointing task and four experimental conditions, demonstrating that the simulated fatigue data matches human fatigue data. We also compare two effort models: 1) instantaneous joint torques commonly used in computer animation and robotics, and 2) the recent Three Compartment Controller (3CC-) model from biomechanical literature. 3CC- yields movements that are both more efficient and relaxed, whereas with instantaneous joint torques, the RL agent can easily generate movements that are quickly tiring or only reach the targets slowly and inaccurately. Our work demonstrates that deep RL combined with the 3CC- provides a viable tool for predicting both interaction movements and user experiencein silico, without users. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|