AI for Unmanned Systems

This repository contains my COMPE 696 AI for Unmanned Systems programming assignments, covering neural-network modeling, PID control for quadrotors, and path-planning algorithms.

Assignments

• Assignment 1 – Neural Networks on MNIST
  Python code implementing:

  • Two simple neural nets with layer sizes (512, 10) and (1024, 10)
  • Two deep nets with layer sizes [1000, 100, 10] and [1024, 512, 256, 128, 10]
    Trains each on the MNIST dataset, logs training vs. testing accuracy, and plots cost vs. epoch curves for four cases :contentReference[oaicite:0]{index=0}.

• Assignment 2 – PID Controller Tuning for UAV
  Python script (drone_pid.py and helpers) that:

  • Implements a PID controller for a quadrotor in X, Y, and Z axes
  • Tunes proportional, integral, and derivative gains to minimize overshoot and settling time
  • Generates convergence plots and discusses simulator setup/debugging challenges :contentReference[oaicite:1]{index=1}.

• Assignment 3 – Grid-Based Path Planning
  Python implementations of:

  • Dijkstra’s algorithm for uniform-cost search
  • A* with an admissible heuristic for faster goal-directed search
    Visualizes obstacle maps, overlays computed shortest paths, and compares node expansions and runtime trade-offs :contentReference[oaicite:2]{index=2}.