Gde where does blockblock come from7/30/2023 ![]() ![]() ![]() The connectors connect the inputs and outputs of nodes. ![]() Note that box and multi detectors will also activate if they are below ground level. ![]() When a screen node is over an obstacle of the matching type, its output will turn on. Screen nodes are the "eyes" of the network: they can tell the network what is currently happening in the game. These have three types: box detectors (blue), triangle detectors (yellow) and multi detectors (green). The inputs of nodes act as boolean OR gates: if any of the connectors connected to that input are on, that input will be on.Īn exception is the output node (the rightmost in the network graph), which needs at least one of its inputs to be on, acting as a boolean OR gate. The nodes act as boolean AND gates: if both of their inputs are on, the output will be on, otherwise the output wil be off (the node accordingly turns either green or red). Each normal node has two inputs: top and bottom. Normal nodes are found in the main network area (with alternating light grey and dark grey columns). Both have different varieties and can be mutated in diffirent ways. The creatures in GDE are composed of nodes (small squares) and connectors (lines). By repeating this process, the creatures (networks) often "evolve", making better progress in the level. The remaining half of the creatures is then cloned - one clone of each surviving creature - and the clones are randomly mutated. Creatures with equal fitness are ordered randomly to increase the variety of different creatures.Īt the end of each generation, half of the creatures are "killed" to make room for new ones. When all creatures in a generation have been simulated, they are sorted by fitness - from best to worst - and then "killed" randomly: the better fitness a creature has, the better is the chance that it survives and vice versa (this relies on a cube function that is taken from carykh's evolutionMATH2 (link at the top)). Based on how far (to the right) the boxes make it, a fitness score is assigned to each creature (equal to the X position of the player when it died). Each creature is simulated independently from the others in the same level. It uses a simple evolution structure: there are generations, which consist of creatures, which consist of nodes and connectors.Įach generation has a certain amount of creatures (currently 300). ( Passing below a (non-flipped) triangle or sliding from a box to the tip of a triangle will also kill the player. Coming into contact with them from any angle, at any velocity will "kill" the cube. However, colliding with the boxes from the side or from below will "kill" the cube. Boxes act like platforms - the cube can slide freely on them and jump off them. Obstructing the player's path there are two types of obstacles: boxes and triangles. Every jump is always of the same height - attempting to jump (or hold the jump button etc.) while in the air has no effect on the trajectory of the player. The cube can jump when it is on the ground or on a box. The player only has one way of control over the cube: jumping. The game consists of a square (the player), which is moving to the right at a constant speed. GDE tries to simulate computer evolution in a simple platformer-like (2D) game. It uses "evolving" simple neural networks to do so.Ī level editor for GDE can be found here. (GDE is not an abbreviation in this project, just a name)) tries to simulate computer evolution in a simple platformer-like game. GDE (or "Project GDE", "Project: GDE" etc. ![]()
0 Comments
Leave a Reply.AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |