2026 ASEE Annual Conference & Exposition

Introductory robotics courses often lean on simulation environments, yet many students encounter a confidence cliff when their code meets real actuators. The bottlenecks are rarely the algorithms themselves, but the integration layers—drivers, timing, calibration, network discovery, controller behavior—and the pace of change across Robot Operating System 2(ROS 2) and vendor stacks. Although ROS 2 is widely adopted in academia and industry, evidence remains limited on whether targeted, low-overhead hardware exemplars improve self-efficacy relative to simulation-only curricula.

We describe a blended lab model that retains simulation while introducing three concrete platforms chosen for breadth and logistics: the Kinova Gen3 Lite manipulator (motion planning and planning scene management); the Turtlebot 4 (mapping/navigation), and Crazyflie micro-UAVs (introductory control). Reproducibility is enforced through prebuilt Docker images and standardized launch/bring-up checklists to mitigate ROS/library drift and machine-to-machine variability. The lab sequence scaffolds from ROS2 introspection and Python nodes to manipulation and mobile tasks, emphasizing safe execution, debuggability, and pathways from RViz/Gazebo to hardware.

A central design tension is guidance versus autonomy; providing enough structure to succeed on physical systems without reducing labs to templates. Course materials have evolved from lightly adapted tutorials to modular lab packets with checkpoints, troubleshooting prompts, and recovery steps. This study involves a cohort comparison between a prior simulation-only offering and the current hardware-integrated term. Pre/post Likert instruments on ROS confidence and debugging autonomy, open-ended student reflections, TA debriefs, and artifact measures (e.g. completion rates, iteration counts, repository practices) will be used to assess efficacy and compare the two robotics integrated labs (with and mostly without hardware).

This paper reports student feedback, performance indications, and the practical benefits and limits of incorporating hardware into an introductory setting. We distill the lessons learned—common sim-to-real pitfalls, containerization trade-offs, safety and planning-scene practices, and strategies for scaffolding without over-prescription—and situate findings within prior work on ROS pedagogy and self-efficacy. We also share reusable assets (container images, launch recipes, modular lab packers) and cost/space considerations to help programs of varying sizes adopt hardware components responsibly. Quantitative results from the ongoing term will be presented alongside qualitative themes to inform future course design.