Swarmify is a simple swarm, motion planning, and collision avoidance simulation based on a potential field method.
It comes with a CPU, a GPU, and a combined implementation.Tested with 2019.4.40f1 and 2021.3.6f1Swarmify is a simple swarm, motion planning, and collision avoidance simulation based on a potential field method.The idea of this project is to provide a basic easy-to-integrate and easy-to-use swarm manipulation toolbox.Check out readme! Collision layer/mask setup is necessary.You can test an earlier version here:To Swarmify Web DemoSwarmify comes with different computation strategies. Depending on your project and target platform you might choose one or the other.Autonomous: Uses Unity's physics engine to filter the distance and compute everything else on the CPU.Centralized CPU: Everything is done on the CPU.Centralized GPU: Everything is fed to a compute shader and the calculations are done there. This leads to a significant performance improvement (however, this is implemented rather crude at this point).Without going into detail. Every agent (Follower) has its own sensor range. Within that, it reacts to other agents by moving away from them. Also, every agent has a target (Leader) that it tries to reach. In other words. The movement directions results of the sum of all repulsive parts (other agents) and the attractive part (target). Properly set up and tweaked this leads to pretty fluent and also collision-free motion/path planning.
