Nowadays, computational thinking (CT) is considered a skill to be taught. While educational robotics (ER) appears promising to foster students’ CT, operational CT frameworks and explicit guidance in ER are lacking. This thesis thus proposes to understand how teachers can implement ER learning activities in order to develop students’ CT competences, and produced two concrete contributions:
1) The creative and computational problem solving (CCPS) model, an operational ER-CT framework that was developed to design ER-CT learning activities, and more specifically the associated intervention and assessment methods, by rendering the CT strategies of students observable.
2) The design and analysis of two complete ER learning activities to promote the full range of CT competences based on the CCPS model, including the artefacts, assessment, and intervention methods.
Both were leveraged in a multi-stage translational research process to investigate how to mediate teach- ing with ER to foster CT. First, 43 teachers following an ER teacher training program were probed regarding their attitude towards ER, which helped identify two distinct profiles of teachers: “pioneers” and “followers”. While the former may implement ER and CT on their own, the latter need resources and guidelines to do so. The analysis highlighted the need for ER activities fostering 21st-century skills. The analysis also showed that as long as teachers can benefit from ER training and take the time to appropriate the robotic artefact (i.e., the first stage of instrumentation), teachers’ attitude toward ER is positive. However, to pass the second stage of instrumentation and thus organize the conditions for developing the students’ CT competence with the robots, ER must be perceived as usable. For this, teachers report needing material resources and time. Since the acceptability of ER is correlated with its usability, work should focus on improving ER usability to increase its acceptability.
The next step thus entailed the design and setup of the two complete ER learning activities, which the CCPS model assessed. The intervention methods targeted the introduction of delays in students’ thought process, once by preventing access to the programming interface and once by having a delay between execution and the visualization of the results of said execution. Results showed success in fostering students’ CT as they went through all the phases involved in the model. Subsequently, an evaluation of teachers’ acceptance of the provided learning activities, intervention and assessment methods was conducted with 334 teachers. Indeed, to orchestrate an ER learning activity for CT, the teachers need time and training to carry out the two instrumentation stages. Results showed that while teachers perceived the utility of the methods and resources provided, it was still unclear how the use of the artefacts (usability) related to students’ strategies and thus CT competences. Hence, teachers need more time for the second stage of instrumentation.
More work must be done, on more extended training periods, to provide explicit guidelines that would help ensure that all teachers and not just pioneers can employ ER to foster CT competences in their classrooms. More explicit guidelines may thus further enable the CCPS model to be a tool for teachers’ instrumentation of CT in ER learning activities in terms of design, intervention and assessment.

Imported from EPFL Memento