Conference proceeding
Predicting Mid-Air Interaction Movements and Fatigue Using Deep Reinforcement Learning
PROCEEDINGS OF THE 2020 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI'20), pp.1-13
01/01/2020
DOI: 10.1145/3313831.3376701
Abstract
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.
Details
- Title: Subtitle
- Predicting Mid-Air Interaction Movements and Fatigue Using Deep Reinforcement Learning
- Creators
- Noshaba Cheema - German Research Centre for Artificial IntelligenceLaura A. Frey-Law - University of IowaKourosh Naderi - Aalto UniversityJaakko Lehtinen - Aalto UniversityPhilipp Slusallek - German Research Centre for Artificial IntelligencePerttu Hamalainen - Aalto UniversityACM
- Resource Type
- Conference proceeding
- Publication Details
- PROCEEDINGS OF THE 2020 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI'20), pp.1-13
- DOI
- 10.1145/3313831.3376701
- Publisher
- Assoc Computing Machinery
- Number of pages
- 13
- Grant note
- 299358 / Academy of Finland 01IS18060C / ITEA3 project MOSIM IMPRS-CS doctoral fellowship
- Language
- English
- Date published
- 01/01/2020
- Academic Unit
- Nursing; Physical Therapy and Rehabilitation Science
- Record Identifier
- 9984294947902771
Metrics
24 Record Views