C++

Model Predictive Controller
This project implements an MPC (Model Predictive Controller) to control actuators on a simulated physics engine.

Path divergence, turning, and breaking sharply are all heavily penalized during the optimization process. The project focused on speed and avoiding obstacles. The result is a car that is able to successfully navigate around the track at 65+ mph.

Please check the project Github for more details.

Particle Filter
An stochastic initial guess of a vehicles coordinates on a semi-detailed map are first created. As the vehicle moves around the environment, our stochastic representation converges by comparing sensor input data and the vehicles current coordinates belief to the semi-detailed map. The best guess of the vehicles position is represented as a blue circle

Please check the project Github for more details.

Unscented Kalman Filter
This project Uses simulated Lidar and Radar data to track a bicycle from the perspective of a moving vehicle. This filter used a CTRV model (Constant turn rate, constant velocity) in order to predict the trajectory of the bike. It has a higher efficacy for objects moving in a circular motion when compared to an Extended Kalman filter.

Please check the project Github for more details.

Extended Kalman Filter
This filter used a CV model (Constant velocity) in order to predict the trajectory of the bike. Due to the curvature of the trajectory taken by the bike, this model has a lower accuracy (Root-Mean-Square-Error) than the Extended Kalman Filter. This filter however, is less computationally expensive.

Please check the project Github for more details.