A new class of distributed, autonomous systems is emerging, capable of exploiting multimodal distributed and networked spatial and temporal data (at significantly larger scales). A renaissance autonomy engineer requires proficiency in both traditional engineering concepts as well as a systems engineering skillset for implementing the ensuing complex systems.
In this paper, we describe goals, development and first offering of a scaffolded course: “AuE 893 Autonomy: Science and Systems” to begin addressing this goal. Geared towards graduate engineering students, with limited prior exposure, the course complements the concepts from traditional courses (on mobile-robotics) with experiential hands-on system-integration efforts (building on the F1tenth.org kits).
The staged course structure initially builds upon open-source Robotics Operating System (ROS) tutorials on simulated systems (Gazebo/RViz) with networked communication; Hardware-in-the-loop realization (with a Turtlebot platform) then aids the exploration (and reinforcement) of autonomy concepts. The course culminates in a final-project comprising performance testing with student-team integrated scaled Autonomous Remote Control cars (based on the F1tenth.org parts-list). All three student teams were successful in navigating around a closed racecourse at speeds of 10–15 miles per hour, using Simultaneous Localization and Mapping (SLAM) for situational awareness and obstacle-avoidance. We conclude with discussion of lessons-learnt and opportunities for future improvement.