What I did

Multi-Lane Navigation
AI racers are capable of switching between multiple lanes along a spline-based track, allowing for more realistic and strategic movement.

Path Decision Logic
When an AI collides with a designated spline trigger box, it:

• Loops through available path options

• Randomly selects a valid path to follow

• Ensures variation in racer behavior across laps
  
  

Steering and Speed Calculation
Each AI racer dynamically adjusts:

• Steering direction based on target lane

• Speed to match curvature and lane changes
  
  

Lane Bound Detection
The system calculates the maximum allowed lane width for each AI:

• Uses a tangent vector from the AI position to the spline center

• Computes lane width based on spline radius at that segment

• Prevents AI from steering outside valid bounds

Forward Box Cast
Each AI racer performs a forward-facing box cast to detect potential collisions with track obstacles or other racers.

Impact Feedback

• Green: no collision detected

• Red: collision detected
  This visualization helps debug avoidance decisions in real time.

Collision Analysis
On impact:

• The system stores both the impact point and the impact normal

• A direction vector from the AI to the impact point is calculated

Side Detection via Cross Product
Using the cross product between the AI’s forward vector and the impact direction, the system determines whether the obstacle is on the left or right side.

Smooth Steering via Interpolation
Based on the side, the AI smoothly steers away from the obstacle over time using interpolation for natural motion.

Reflection Vector Steering
A reflection vector is calculated using the actor’s forward vector and the impact normal. This vector is used to determine an alternate safe direction.

Dynamic Road Positioning
The reflection direction helps decide which side of the road the AI should steer toward, ensuring consistent avoidance behavior across varied track layouts.