Embodied AI Challenge: Autonomous Miniature Race Car Edition

Abstract

Build an autonomous miniature race car according to the rules of the NXP Cup. As a team of about three students from different programs, you will design and build your car and the software stack to compete against other teams in the course. Top teams can then proceed to the actual NXP Cup. The course provides a challenge-based setting in which students develop an autonomous mobile robot integrating embedded computing, camera-based perception, real-time control, and mechanical components. Students define their own timeline, deliverable, etc., on a topic where intelligence emerges from the interaction between learning algorithms, physical hardware, and the environment.

This course is primarily aimed at Master students.

Learning Objectives

By the end of the course, students will be able to:

• Design an embodied-AI system: Develop an autonomous vehicle that closes the perception–decision–action loop by sensing its environment, interpreting sensor data, selecting appropriate actions, and controlling steering and speed in real time.
• Integrate hardware, software, and AI components: Design and implement a complete embedded software stack that combines sensors and cameras, perception or machine-learning algorithms, control functions, real-time microcontrollers, motor drivers, and the physical vehicle platform.
• Experimentally optimize a cyber-physical system: Systematically test, measure, debug, and improve the vehicle with respect to accuracy, speed, robustness, latency, and resource consumption, while evaluating trade-offs between performance and operational reliability.
• Work effectively in an interdisciplinary engineering team: Collaborate in a team of about three students from different study programs, distribute responsibilities according to complementary expertise, integrate independently developed subsystems, document design decisions, and jointly present and defend the final system.