Matthew Loveday

For my Artificial Intelligence in Games University module I was tasked with developing a tank which could complete an arena, the arena itself has two invincible guardians and two mandatory pickups.

To complete this task I implemented three core features into the tank, these were:

1. Steering Behaviours
2. Directed Recursion Search (pathfinding)
3. Finite State Machine (decision making)

The tank is capable of completing the whole course while maintaining it's health due to the variety of behaviours and decision making techniques used, for example when it encounters one of the guardians it seeks a hiding spot ensuring it's safety. For this assignment I was awared 100% of the marks.

Before this module I had never had any serious attempt at making a competent AI which was able to complete complex tasks using complex behaviours, the most I had done previously was simple pathfinding. Now I am able to create AI using finite state machines, behaviour trees, steering behaviours and much more.

If I was going to do anything differently it would be to focus more on adding states, the three states work but it lacks granularity in some cases, for example there is a simple flee state where as it could be better to have a flee state and a hide state as distinct behaviours.