Tourney of Eradorn

Trailer

Overview

In Tourney of Eradorn players compete against AI enemies for the title of "Champion of Eradorn". Throughout the tournament you may collect powerful weapons, armor, and totems to increase your strength. It features directional attack melee combat with enemy AI that adapts to your playstyle.

Tourney of Eradorn is the first game I ever released.

My Role

Sole developer and designer.

Inspiration

When the GearVR was released I bought one on a whim to test out the emerging technology of VR headsets. I was blown away by how immersive it was, and was convinced in the future of the technology. There were no good sword fighting games available on the Oculus Store, and I really wanted to play one.

Battle royales the FOTM and I found the technical challenge of building battle royale enemy AI to be very interesting.

Design

• Large map with objectives and procedurally placed items
• Battle Royale "wall of death"
• Boss fights with unique mechanics
• Directional attack melee combat
• Combo system
• Enemy AI that adapts to playstyle
• Powerups, Weapons with unique abilities, armor upgrades

Outcome

I released Tourney of Eradorn on the Oculus Store for GearVR and Oculus GO headsets in 2019. The game was purchased 1000s of times.

Tourney of eradorn was completed after 3 years of development in my spare time. Being my first game it is not an accurate measure of my abilities, but it taught me value of pesistence and what it takes to release a game.

Featured Work

NGram Predictors

Players/AI can perform attacks/block from three directions: left, right, middle. Blocking from the direction of an incoming attack with disable the attacker for a short period of time, providing an opening. Weapons have combos that do bonus damage/apply status affects when performed (ex: epic slash left, right, middle).

I implemented the enemy combat AI to use Ngram predictors to determine what their next attack should be (same algorithm used for auto complete in text messages). Each time the player attacks their attack is stored within a tree that tracks all previous sequences of attacks. This creates a history of attack sequences the player has performed, and provides an effiencent lookup structure for predicting what the next most likely attack will be. When evaluating their next action the AI queries the tree to determine the next most likely attack, and chooses an approriate counter swing/block. To adjust difficultly noise is added to give some variance in predictions.

This works extremely well for predicting combos, and the prediction tree is persisted through play sessions to create a profile of the player's "playstle".

Goal Action Oriented Programming (GOAP)

During tournament rounds the player is placed into a battle royale with AI enemies. The AI utilized Goal Action Oriented Programming to generate plans in responds to current game state. Each enemy has a Personality that determines the costs of performing certain actions when travesring the action space. The weight of these actions and the current game state determines what the enemy's current Plan. Plans are reevaulated at a configurable rate.

• Ex: timid enemy searches for gear
• Ex: aggressive searches for other combatants to attack

Diagetic UI

At the time VR games were very new, and I felt most games were not adapting their UI to the medium. I kept as much UI in world as possible for better immersion.

• Inventory as a rolled out pack with your items on the ground
• Physical map you held and looked at