Sword Snow Stride

Enemy AI Design
Project Overview
Upcoming UE5 PvPvE action RPG

Designed and deployed the AI behavior using Dynamic AI System (DAS), designed and assembled the AI combat system using Gameplay Ability System (GAS) with skill set provided by the combat design team.
Role
Level Designer
|
AI Combat System Design
AI Behavior Design
1. Determine Appropriate Enemy AI Allocation
The allocation of enemy AI is initially determined by considering the size of levels and resource quotas, which encompass factors such as loot drops, chests, and more.
2. Design Enemy AI Patrol Routes
After determining the allocation, we proceed to design enemy AI patrol routes based on players' flow outside the levels, the indoor and outdoor layouts, and resource distribution within the levels. This ensures a consistent balance between [1] the efforts and risk factors the players invest, encompassing both PvE challenges and the potential exposure leading to PvP engagements, and [2] the rewards they receive, regardless of their entry and exit points within the levels.
3. Ensure A Diverse Experience (within the level and among levels)
The next objective is to ensure a diverse and engaging player experience, which stands as one of the central pillars of our level design. This goal is chiefly accomplished through two key strategies:
(1) Creating Distinctiveness: We strive to introduce variations between different levels within the map and among different locations within each level. We meticulously tailor each encounter, ensuring unique enemy compositions and placements. For instance, while facing an indoor challenge in Level A, players may confront a melee enemy and an elite one, while in Level B's outdoor section, they might encounter two ranged enemies and an elite one. Furthermore, we differentiate patrol routes, enemies march side by side in some areas, but follow an zigzag pattern in others.
(2) Weather System Integration: As a PvPvE game, we prioritize the replayability of our levels. To achieve this, we integrate our levels with a dynamic weather system. By doing so, players can explore distinct strategies under varying weather conditions, leading to diverse critical paths and unique experiences in each game. For example, within a level set in a inn, elite enemy may showcase their kungfu on an outdoor stage during a clear night, while he opt to guard the indoors chest himself during rainy weather. Similarly, in a temple-themed level, enemies might engage in rainmaking rituals indoors on sunny days, but come outside to celebrate and cheer during rainfall.
Through these meticulous designs, we aim to offer players ever-changing experiences within our levels. It's essential to ensure that the patrol routes we design here also meet the requirements we set in the "Design Enemy AI Patrol Routes" section above. The diverse behaviors of enemies also imposes higher demands on our plausibility design, which will be discussed in the next section.
4. Design Plausibility
The final step is to ensure that enemy AI behavior feels believable and immersive. We carefully address questions like why an enemy appears at a specific time, who they are, and their purpose in that spot. This is a collaborative effort involving game writers and the art team. For each level, we create multiple behavior scenarios, including story settings, character backgrounds, required actions, and necessary props in the scenes. Once the primitive design is in place, we closely cooperate with the level artists to ensure that the visual design aligns with the intended atmosphere. Concurrently, we work alongside game writers to fine-tune the narrative and dialogues. After finalizing these aspects, animators and modelers provide us with the animations and props. Through this collaborative approach, we achieve enemy AI behavior that feels authentic and seamlessly integrated into the overall consistent level experience.
AI Behavior Design - Early Stage Draft
Here is one early stage draft that illustrates how the level layout and resource distribution shape the critical paths, impacting the allocation of enemy AI.
AI Behavior Design - Refined Draft
Once the allocation is preliminarily established, we proceed with the concrete design of AI behavior. Below are some of the levels where I'm in charge of the enemy AI behavior design.
Legend
AI Behavior - Inn
Clear
Rainny
1st Floor
Yard
Yard
1st Floor
2nd Floor
2nd Floor