Abstract
A study of long-term interaction with the robot embodiment of the companion called Sarah was conducted during the summer of 2012. The aim of the study was to see long-term implications when the robot embodiment was in a natural setting.
The robot interacted with 5 participants for 3 weeks in a office environment running continuously. Analysis of such a longterm experiment is a big challenge. Current robotics research has mostly addressed long-term interaction as a re- peated interaction on a fixed task (e.g.(Leite et al. 2012)), but here we investigated a continuous interaction of 3 weeks. The challenge is not only to evaluate the model implemented on the robot, but also the resulting behaviour of the robot. A method to evaluate the resulting behaviour is to evaluate the loop between human and robot. Hence, the idea is to evalu- ate the interplay presented in the interaction of human and robot (e.g (Lohan et al. 2012)).
The embodiment of a system like a robot in social situa- tions is defined as not only dependent on its own sensory- motor experiences and capabilities but also on the environ- mental changes, caused by social constraints (Dautenhahn, Ogden, and Quick 2002). Thus, moving robots into social environments needs to be able to take their surroundings and the rules given by these surroundings into account. This is why from a robotic perspective, the interplay between its be- haviour and the behaviour of its interaction partner(s), needs to be considered carefully.
Long-term social interaction between robot and human(s) creates very complex evaluation issues. Methodologies used in developmental psychology suggests that it is difficult to create a quantitative strategy to evaluate the long-term evo- lution of interactions. Hence, most research uses a sampling procedure, i.e. a experience-sampling procedure (Steiger et al. 1999).
A further problem that needs to be addressed in evaluating long-term interactions is one of big-heterogeneous data that results from it. This dataset comprises both video capture and system log files. To scale down the search space in the data, key-points in the interaction must be identified. There- fore, in this paper we consider the use of energy consump- tion of the robot to predict these key-points.
The robot interacted with 5 participants for 3 weeks in a office environment running continuously. Analysis of such a longterm experiment is a big challenge. Current robotics research has mostly addressed long-term interaction as a re- peated interaction on a fixed task (e.g.(Leite et al. 2012)), but here we investigated a continuous interaction of 3 weeks. The challenge is not only to evaluate the model implemented on the robot, but also the resulting behaviour of the robot. A method to evaluate the resulting behaviour is to evaluate the loop between human and robot. Hence, the idea is to evalu- ate the interplay presented in the interaction of human and robot (e.g (Lohan et al. 2012)).
The embodiment of a system like a robot in social situa- tions is defined as not only dependent on its own sensory- motor experiences and capabilities but also on the environ- mental changes, caused by social constraints (Dautenhahn, Ogden, and Quick 2002). Thus, moving robots into social environments needs to be able to take their surroundings and the rules given by these surroundings into account. This is why from a robotic perspective, the interplay between its be- haviour and the behaviour of its interaction partner(s), needs to be considered carefully.
Long-term social interaction between robot and human(s) creates very complex evaluation issues. Methodologies used in developmental psychology suggests that it is difficult to create a quantitative strategy to evaluate the long-term evo- lution of interactions. Hence, most research uses a sampling procedure, i.e. a experience-sampling procedure (Steiger et al. 1999).
A further problem that needs to be addressed in evaluating long-term interactions is one of big-heterogeneous data that results from it. This dataset comprises both video capture and system log files. To scale down the search space in the data, key-points in the interaction must be identified. There- fore, in this paper we consider the use of energy consump- tion of the robot to predict these key-points.
| Original language | English |
|---|---|
| Publication status | Published - 2014 |
| Event | Artificial Intelligence for Human-Robot Interaction: AAAI 2014 Fall Symposium Series - Arlington, VA, United States Duration: 13 Nov 2014 → 15 Nov 2014 |
Conference
| Conference | Artificial Intelligence for Human-Robot Interaction |
|---|---|
| Abbreviated title | AI-HRI 2014 |
| Country/Territory | United States |
| City | Arlington, VA |
| Period | 13/11/14 → 15/11/14 |
