Windstriders

Project Management Skills

For the first semester of development, pre-production and early production of the project, I was the producer for the project and handled team coordination, task delegation, client communications, leading scrums, deliverable submissions, and emergency design decisions. I ensured everyone on the team shared a united vision of the project, both the client and course teaching staff were happy with the project and it's progress, and that development was happening timely and to an adequate standard. This was done through effective utilisation of a number of tools, including Github to ensure we could keep all work centralised to one project, Google Calendar to confirm team availabilities and schedule meetings with the client across time zones, and Trello to track all the tasks being worked on, as well as track the product backlog and create burndown charts.

Handling pivots in design and deciding if they're necessary was another responsibility of mine as the Producer, and one of the first major pivots was in the rope system I personally was developing. The initial ropes were dynamically generated and utilised physics and hinge joints to create the physics of the rope, with a spline overlaid for the visual elements to create a cohesive appearance, however there were numerous issues with how the rope interacted with the boat when physics was enabled for both. As well as this, the ropes could be distracting and hard to find, making the player experience worse. As a result of this, I made the decision to pivot away from physics based ropes, slightly decreasing the realism of the project to improve the play experience and make the game better overall.

Dynamic AI Development

A major component to Windstriders is the Race Mode, where you can race in a fleet of sailors on tracks inspired by real Olympic sailing layouts and locations. To make this possible, we had to develop an AI system capable of navigating the course while accounting for the various factors that influence how an individual should be sailing, including wind strength, direction, proximity to other sailors, and the ideal sail shape/tightness for the conditions.

Developing this AI was completely my responsibility, and after some initial research I decided to utilise a Goal Oriented Action Planning (GOAP) system to drive the AI's actions. The two main reasons for this were that the AI boats are required to operate the same way as the player boat by directly controlling the various ropes and rudder, as well as the dynamic, emergent nature that GOAP offers to an AI system. By being able to plan a series of actions independently based on what it believes to be the best option for achieving the current goal, minor differences between each specific sailboat can lead to very different behaviours, with some boats acting far more aggressively with others playing passive and waiting for the perfect moment to shoot around a buoy for an overtake.

Utilising GOAP AI also allowed for a much greater scale of AI much easier than a more manual system. During testing, the AI functioned with 100+ sailboats on the course at once, although the course wasn't designed for this number of boats and so the behaviour became very chaotic. However, the mode was intended to be played in fleets closer to 16-24 ships, as seen below, which the AI performed much better in