PD vs. NMPC for Jetracer Car Trajectory Tracking | Uncertain Experiments

The video shows the experimental results obtained by comparing a classical PD controller with a Nonlinear Model Predictive Controller (NMPC) for a Jetracer steering car trajectory tracking.
Two non-linear trajectories have been considered: a circle and a Bernoulli's lemniscate trajectory.
All the tests consider that halfway through the experiment, the state's measurements are not lost, i.e., not updated for 5 seconds.
On one hand, with the PD controller, the robot highly deviates drawing circle.
On the other hand, with the NMPC, the steering car stays close to the reference, and once the measurements are available again, the agent corrects its trajectory to come back on its way.

Видео PD vs. NMPC for Jetracer Car Trajectory Tracking | Uncertain Experiments канала Autonomous Vehicles