Human-Swarm Interaction Robotics as Context for Training Diverse Undergraduate Researchers.

Our objective in this Evidence-Based Practice (EBP) paper is to describe an innovative approach and the impact of that approach on undergraduates engaged in engineering and computer science research experience involving robotics swarms. While critiques of a narrow reading of EBP approaches rightly assert inapplicability when applied as a mandate to educational practitioners [1], our goal here is to contribute to the evidence base supporting our educational practitionerdriven pedagogy thereby expanding the definition and applicability of EBP in education. More recent approaches to EBP in disciplines aligned with the learning sciences have made it clear that a broad range of evidentiary warrant (i.e., beyond quantitative measures in randomized control trials), integration of cognitive learning theory with human developmental theory, and both practitioner and learner input are critical to developing an evidence base appropriate to guide educational practice [2], [3]. We seek to help undergraduate students develop into independent researchers through challenging them with real-world problems utilizing human-swarm interactions as both research context and pedagogical model for how to engage students in undergraduate research. We describe our program and the sequential mixed methods evaluation informing quality improvement and summative evidence of impact on participants. Our undergraduate researchers explore innovative ways of interacting with collectively emergent systems behavior without directly controlling individual building blocks of swarms, and we utilize our understanding of human-swarm interaction as a model for organizing their research experience. Students gain first-hand knowledge of the nonlinearity and robustness of robot swarm behaviors through observations and interactions during hands-on experiments. We give students ownership of engineering problems formulated as challenges their "swarm" must solve. With this active learning approach, students' creativities are stimulated, and they become more confident, comfortable, and competent in solving complex robotics problems [4], [5]. We utilized a sequential mixed methods evaluation design for triangulation, complementarity, and development [6], [7]. This allowed us to leverage analysis of collected data to inform development of later data sources (i.e., development), as well as to examine the degree of overlap (i.e., triangulation) and independent contribution (i.e., complementarity) across quantitative and qualitative data sources. Data sources included a pre-post survey and focus groups during week 4 and week 8 of our 10-week program. These data sources allowed quantitative pre-post change analysis to be triangulated with participant self-assessment of impact and complemented with participant description of activities and perspectives that facilitated such impact as well as suggestions for improvement. Participants described positive impacts of the autonomy-supportive structure and authentic nature of the program in all areas and most strongly in Research Skills development. Participant recommendations centered on better communication, increased direct support from mentors, and focus on leadership, careers/graduate school, and scientist identity development. We recommend consideration of literature on cognitive apprenticeship in communities of practice organized around research groups [8] to inform projects such as this. [ABSTRACT FROM AUTHOR]