First Devlog!

Devlog 1: The Sheep Brain And General Setup

“Great minds flock alike!” - the sheep in our game probably. This week we began working on the general setup of the game, adding various 3d assets (sheep, border collie, level 1 ranch, background elements), letting the dog run around annoying the sheep, work on the goal state of the game and of course the ingenious sheep AI. 

In case you are unfamiliar with the game, the goal is to herd a flock of sheep home after a storm. You must collect the strays and keep the group together and make sure you get them home safe while fending off any challenge the game and its environment throws at you along the way.

The Sheep

Initially the sheep started off as Unity NavMesh agents, which may have seemed like a simple obvious choice. However a sheep in a flock doesn’t act like an entity with a set destination in mind, rather it reacts to its fellow sheep, the dog and any other obstacles scattered around.  So we replaced this with a genius steering behaviour (similar to the Reynolds flocking approach). Here’s a breakdown of how it works- 

• Cohesion: Each sheep obtains the locations of its 3 nearest neighbours within 8 metres and approximates the average/center and is pulled towards it.
• Separation: sheep push away from the same neighbours if they get too close (avoids a sheep stampede)
• Dog avoidance: When the dog gets too close to a sheep it gets pushed directly away with the magnitude of this vector increasing the closer the dog gets. 
• Wander: When the dog is no longer bullying the sheep, they have a chance to ‘wander’, i.e head in a random direction. 
• Obstacle Avoidance: Each sheep checks ahead with a single sensor and if blocked, fires 7 probes that fan out, finding an alternate path.

Additionally we worked on a scatter behaviour, which can be paired with a barking mechanic. Scatter essentially reacts to a signal from a source position (like a dog barking) and makes the sheep pick a random angle away from it and accelerate in that direction for a few seconds. Compared to games like Twilight princess where we took some inspiration, some from our sheep herding feels a lot more realistic, with often smaller sub-flocks being formed in the simulation. The video below demonstrates a top down view of the sheep grazing over time.

Goal State

For the purposes of the game jam, we implemented a simple goal mechanism that triggers a win condition when enough sheep are inside the pen.

Our future idea for the “goal” is to have one at the end of each level. It would act as the objective point where the player and sheep go to transition to the next level. Instead of requiring a number of sheep to be present, we want future builds to have a more lenient system; we want it to be possible for players to govern how many sheep they manage to herd to the end of each level, which may influence game endings.

Level Design

We want to make levels that express the flocking behaviours in new ways. We plan to utilize obstacles that the sheep runs into, creating situations including a split flock that the player has to adjust for.

Music and Art

Our art team created several assets for us to implement into the demo scene, grass textures, trees, fences, sheds, rocks and other obstacles (We even made a skateboard version of the border collie for mysterious reasons). 

For SFX our team compiled many barking sounds both from online sources and their own pets, which we are working on integrating smoothly with bark mechanics. They even produced some ambient background tracks that could be used to set the vibe. 

For our next steps, we plan on fully integrating barking mechanics, improving the sheep AI, including different variants of sheep and decorating our starting levels. Stay tuned to see how our game (and sheep) keep evolving!