Weston Smith's repositories
PID-Control
Implemented PID controller with twiddle algorithm to continuously tune hyperparameter in order to autonomously drive a vehicle around a track given the vehicle's cross-track error.
Backyard-Flyer
Built a flight controller that utilizes event-driven programming to autonomously fly a quadcopter through a pre-defined path.
CarND-Capstone
This is the final project in Udacity's Self-Driving Car Engineer Nanodegree where we will implement ROS nodes to control Carla - Udacity's self-driving car.
Motion-Planning
This project implements path planning techniques as part of Udacity's Flying Car Nanodegree program.
Advanced-Lane-Lines
Applied signal processing to dash-cam video feed to detect lane lines on the road and used numerical methods to derive approximate real-world measurements of the lane lines.
Behavioral-Cloning
Built a convolutional neural network using Keras that learns to autonomously drive a vehicle around a track through observation.
Extended-Kalman-Filters
Implemented sensor fusion algorithm using an extended Kalman filter that tracks nearby moving objects using RADAR and LIDAR measurements.
Finding-Lane-Lines
Implemented canny edge detection and hough transform in lane-finding pipeline to detect lane lines in video streams.
Kidnapped-Vehicle
Implemented a particle filter in order to accuractly localize a vehicle using sensor measurements, an initial noisy GPS reading, and a map of the region.
Path-Planning
Implemented a path planner to intelligently navigate a vehicle through a highway environment using trajectory generation, behavioral planning, and prediction of surrounding vehicles.
Quadrotor-Controls
This is the thrid project in Udacity's Flying Car Nanodegree which implements a flight controller for a quadrotor in C++.
Quadrotor-Estimation
This is the final project in Udacity's Flying Car and Autonomous Flight Engineer Nanodegree which covers the estimation portion of a flight controller.
Traffic-Sign-Classifier
Built and trained a convolutional neural network with over 95% accuracy to classify a data set of German traffic signs based on the LaNet-5 architecture.