The project started with an experiment to see if an agent could follow a line and illuminate its path to see obstacles. The light is composed of isovits and only the ones in front of the agent are colored. The experiment went then to make the light 3D to illuminate buildings facades. Many approaches were tried but ultimately failed due to high computing when calculating the intersections between meshes.
Unlike the first iteration where the agent followed a predefined line, the goal now was to achieve full autonomy and design an agent that can complete a course without receiving any guidance and no mater where it is set on the course. The main loop had 7 steps:
- Agent starts loop
- Agent launches lasers to find edges and measure distance
- Agent determines top 3 longest laser lines
- Agent finds a point that is the weighted average among the top 3 longest lines. This is necessary to ensure the agent doesn’t get stuck.
- Using that point, the agent creates a direction vector to follow until the next loop. This is the main component that allows the agent to navigate any track.
- Agent displays light and laser intersections.
- Agent moves to the next point on the direction vector.
- Loop ends
Using this logic the agent can navigate its way on any course.
