Robot education peers in a situated primary school study: Personalisation promotes child learning

Abstract

The benefit of social robots to support child learning in an educational context over an extended period of time is evaluated. Specifically, the effect of personalisation and adaptation of robot social behaviour is assessed. Two autonomous robots were embedded within two matched classrooms of a primary school for a continuous two week period without experimenter supervision to act as learning companions for the children for familiar and novel subjects. Results suggest that while children in both personalised and non-personalised conditions learned, there was increased child learning of a novel subject exhibited when interacting with a robot that personalised its behaviours, with indications that this benefit extended to other class-based performance. Additional evidence was obtained suggesting that there is increased acceptance of the personalised robot peer over a non-personalised version. These results provide the first evidence in support of peer-robot behavioural personalisation having a positive influence on learning when embedded in a learning environment for an extended period of time.

Fig 1. Typical physical setup of the system within the classroom.

The robot, Sandtray—a touchscreen device—and camera setup was located in one corner of the room in which the children had their normal lessons. Interactions took place during normal lesson time. Both classrooms had similar arrangements. Not to scale.

Fig 2. Interaction structure and contents.

(a) structure of each interaction, with five minutes on the collaborative sorting task itself; (b) example of a child engaged in the task with the robot (hardware and classroom setup as shown in Fig 1); (c) two sample image libraries, showing a 3 times-table task, and a stone-age animals task.

Fig 3. Library scores per image library.

Overview of mean scores per library, by condition, error bars are 95% CI: (a) performance in each of the image libraries, see Table 1 for library contents; (b) scores for the first four stone-age image libraries (novel subject): ‘*’ denotes significance at the .05 level.

Fig 4. Child learning performance between conditions.

(a) summary of mean percentage test scores (for pre and post experimental period) for the familiar learning task (times-tables), the novel learning task (the stone age), and the independent task (spelling, for which there was also a mid-experiment test); (b) normalised learning gain exhibited in the familiar, the novel, and the independent learning tasks. Error bars show 95% CI.

Fig 5. End-of-interaction question responses.

(a) end of interaction responses after the second and third interactions to the question “what would you prefer to play with next?”, with “none” recorded if an answer is not given within 30 seconds (multiple choice from: robot, classmates, read a book, play outside, games console, or no answer); (b) box-plots showing child ratings for the four questionnaires (end of bars represent last datum within the 1.5*IQR; circles denote outside values; no outliers): social presence, social support, interest/enjoyment and perceived competence. Crosses indicate the mean, numbers below the bars denote sample size.