Lecture 8 (Collision Detection)

Question 1

What is collision detection?

Answer 1

Detection of an intersection between different objects in a virtual environment

• Prevents objects from just going through each other
• Produce realistic virtual environments by having objects react accordingly when a collision happens

Question 2

What are the collision detection approaches?

Answer 2

• Continuous (a priori)

- Discrete (a posteriori)

Question 3

Describe: Continuous (a priori)

Answer 3

We predict if a collision is going to occur BASED ON THE TRAJECTORIES of the objects.
- i.e. calculate if they will hit before they actually hit

** Is computationally complex, can affect performance

Question 4

What are the drawbacks of Continous (a priori)?

Answer 4

If you have 1 object with 100 triangles, and another one with 200 triangles.

Then there are 100 * 200 = 20,000 different potential intersections to calculate.

• i.e. It is computationally complex, can affect performance

Question 5

Describe: Discrete (a posteriori)

Answer 5

Registers the collision after it has happened.
- i.e. Think of it like a quick rewind in time.

Much more efficient than continuous (a posterior)

Question 6

What are the drawbacks of Discrete (a posteriori)?

Answer 6

Tunnelling may occur.

Question 7

What is Tunnelling?

Answer 7

When a very fast object passes through another object, and the engine doesn’t even get the chance to detect the collision.

Happens with Discrete (a posteriori).

Question 8

What are bounding volumes? Why are they useful? Give examples

Answer 8

Used to further simplify collision detection - don’t need to detect every polygon.

Volumes will enclose the whole object and should be as small as possible for accuracy.

• Axis Aligned Bounding Box (AABB)
• Oriented Bounding Box (OBB)
• Bounding Spheres

Question 9

What is ‘Axis Aligned Bounding Box’ (AABB)?

Answer 9

Bounding box remains oriented WITH RESPECT TO THE MAIN AXES

Question 10

Pros and Cons of Axis Aligned Bounding Box (AABB)?

Answer 10

Pros:

• Very simple to compute
• Translation invariant (never changing)

Cons:

• The box will not change if you move every point of the object by the SAME amount in a given direction, but for other movements, box will no longer remain axis-aligned
• Bounding box needs to be recomputed on every frame

Question 11

What is ‘Oriented Bounding Box’ (OBB)?

Answer 11

Bounding box is oriented WITH RESPECT TO THE OBJECT

Question 12

Pros and Cons of Oriented Bounding Box (OBB)?

Answer 12

Pros:

• Transformation invariant (never changing)
  
  • Translation
  • Reflection
  • Rotation
• Volume is more compact than AABB, therefore there is less empty space

Cons:
- The computation of intersections is, however, much more complex as box orientation is dynamic

Question 13

What are ‘Bounding Spheres’?

Answer 13

Collision detection approach similar to Oriented Bounding Box, but instead of box, we using spheres

Question 14

Pros and Cons of Bounding Spheres?

Answer 14

Pros:

• Transformation invariant
  
  • Translation
  • Reflection
  • Rotation
• Simple collision detection calculation
  
  • Bounding spheres collide if distance between centres of two objects is less than sum of radii

Cons:
- More empty space in the volume

Question 15

What are the two-phase approaches to improve bounding box efficiency?

Answer 15

Broad phase (simple, low cost) and Narrow phase (complex, high cost).

• Broad phase does initial general sweep, if it picks something up, it then activates the Narrow phase to pinpoint the exact collision point.

Question 16

What is the Broad Phase in hierarchical bounding volumes?

Answer 16

This is the first pass on the bounding volumes.

It is intended to identify POTENTIAL collisions, while also marking objects that are definitely not colliding.

Question 17

What is the Narrow Phase in hierarchical bounding volumes?

Answer 17

Based on the identified POTENTIAL collisions from the Broad Phase, EXACT collision tests are then conducted to determine if the object will actually collide or not.

Question 18

Describe Hitboxes and what they are typically used in.

Answer 18

Hitboxes are simplified bounding volumes as they are used to detect one-way collisions, such as bullets hitting a character’s head.

• Typically used in video games for real-time collision detection.