Control and state estimation are essential in the typical electrical and mechanical engineering curriculum. Obtaining a conceptual understanding on these subjects has proven to be difficult however. Moreover, using the learned concepts to provide an integrated solution for a real-world setup is even more challenging. This paper describes a mobile cart project that promotes conceptual and deep learning of (feedback and feedforward) control and state estimation (Kalman filter) in a master course on control theory. The overall goal for the students is to control a LEGO Mindstorms robot such that it follows a predefined path using the robot's encoders and an ultrasonic sensor measuring the distance to two perpendicular walls. Additionally this paper identifies typical challenges students encounter when faced with such an integrated real-life project integrating control and estimation.