Circle line-segment collision detection algorithm?

No one seems to consider projection, am I completely off track here?

Project the vector AC onto AB. The projected vector, AD, gives the new point D.
If the distance between D and C is smaller than (or equal to) R we have an intersection.

Like this:

Community Edit:

For anyone stumbling across this post later and wondering how such an algorithm can be implemented, here is a general implementation written in JavaScript using common vector manipulation functions.

/**
 * Returns the distance from line segment AB to point C
 */
function distanceSegmentToPoint(A, B, C) {
    // Compute vectors AC and AB
    const AC = sub(C, A);
    const AB = sub(B, A);

    // Get point D by taking the projection of AC onto AB then adding the offset of A
    const D = add(proj(AC, AB), A);

    const AD = sub(D, A);
    // D might not be on AB so calculate k of D down AB (aka solve AD = k * AB)
    // We can use either component, but choose larger value to reduce the chance of dividing by zero
    const k = Math.abs(AB.x) > Math.abs(AB.y) ? AD.x / AB.x : AD.y / AB.y;

    // Check if D is off either end of the line segment
    if (k <= 0.0) {
        return Math.sqrt(hypot2(C, A));
    } else if (k >= 1.0) {
        return Math.sqrt(hypot2(C, B));
    }

    return Math.sqrt(hypot2(C, D));
}

For this implementation I used a couple common vector manipulation functions that you are likely to already have provided in whatever environment you might be working in. However if you do not already have these functions available, here is how they can be implemented.

// Define some common functions for working with vectors
const add = (a, b) => ({x: a.x + b.x, y: a.y + b.y});
const sub = (a, b) => ({x: a.x - b.x, y: a.y - b.y});
const dot = (a, b) => a.x * b.x + a.y * b.y;
const hypot2 = (a, b) => dot(sub(a, b), sub(a, b));

// Function for projecting some vector a onto b
function proj(a, b) {
    const k = dot(a, b) / dot(b, b);
    return {x: k * b.x, y: k * b.y};
}

Taking

1. E is the starting point of the ray,
2. L is the end point of the ray,
3. C is the center of sphere you're testing against
4. r is the radius of that sphere

Compute:
d = L - E ( Direction vector of ray, from start to end )
f = E - C ( Vector from center sphere to ray start )

Then the intersection is found by..
Plugging:
P = E + t * d
This is a parametric equation:
P_x = E_x + td_x
P_y = E_y + td_y
into
(x - h)² + (y - k)² = r²
(h,k) = center of circle.

Note: We've simplified the problem to 2D here, the solution we get applies also in 3D

to get:

1. Expand x² - 2xh + h² + y² - 2yk + k² - r² = 0
2. Plug x = e_x + td_x
   y = e_y + td_y
   ( e_x + td_x )² - 2( e_x + td_x )h + h² + ( e_y + td_y )² - 2( e_y + td_y )k + k² - r² = 0
3. Explode e_x² + 2e_xtd_x + t²d_x² - 2e_xh - 2td_xh + h² + e_y² + 2e_ytd_y + t²d_y² - 2e_yk - 2td_yk + k² - r² = 0
4. Group t²( d_x² + d_y² ) + 2t( e_xd_x + e_yd_y - d_xh - d_yk ) + e_x² + e_y² - 2e_xh - 2e_yk + h² + k² - r² = 0
5. Finally, t²( d · d ) + 2t( e · d - d · c ) + e · e - 2( e · c ) + c · c - r² = 0
   Where d is the vector d and · is the dot product.
6. And then, t²( d · d ) + 2t( d · ( e - c ) ) + ( e - c ) · ( e - c ) - r² = 0
7. Letting f = e - c t²( d · d ) + 2t( d · f ) + f · f - r² = 0

So we get:
t² * (d · d) + 2t*( f · d ) + ( f · f - r² ) = 0

So solving the quadratic equation:

float a = d.Dot( d ) ;
float b = 2*f.Dot( d ) ;
float c = f.Dot( f ) - r*r ;

float discriminant = b*b-4*a*c;
if( discriminant < 0 )
{
  // no intersection
}
else
{
  // ray didn't totally miss sphere,
  // so there is a solution to
  // the equation.
  
  discriminant = sqrt( discriminant );

  // either solution may be on or off the ray so need to test both
  // t1 is always the smaller value, because BOTH discriminant and
  // a are nonnegative.
  float t1 = (-b - discriminant)/(2*a);
  float t2 = (-b + discriminant)/(2*a);

  // 3x HIT cases:
  //          -o->             --|-->  |            |  --|->
  // Impale(t1 hit,t2 hit), Poke(t1 hit,t2>1), ExitWound(t1<0, t2 hit), 

  // 3x MISS cases:
  //       ->  o                     o ->              | -> |
  // FallShort (t1>1,t2>1), Past (t1<0,t2<0), CompletelyInside(t1<0, t2>1)
  
  if( t1 >= 0 && t1 <= 1 )
  {
    // t1 is the intersection, and it's closer than t2
    // (since t1 uses -b - discriminant)
    // Impale, Poke
    return true ;
  }

  // here t1 didn't intersect so we are either started
  // inside the sphere or completely past it
  if( t2 >= 0 && t2 <= 1 )
  {
    // ExitWound
    return true ;
  }
  
  // no intn: FallShort, Past, CompletelyInside
  return false ;
}

I would use the algorithm to compute the distance between a point (circle center) and a line (line AB). This can then be used to determine the intersection points of the line with the circle.

Let say we have the points A, B, C. Ax and Ay are the x and y components of the A points. Same for B and C. The scalar R is the circle radius.

This algorithm requires that A, B and C are distinct points and that R is not 0.

Here is the algorithm

// compute the euclidean distance between A and B
LAB = sqrt( (Bx-Ax)²+(By-Ay)² )

// compute the direction vector D from A to B
Dx = (Bx-Ax)/LAB
Dy = (By-Ay)/LAB

// the equation of the line AB is x = Dx*t + Ax, y = Dy*t + Ay with 0 <= t <= LAB.

// compute the distance between the points A and E, where
// E is the point of AB closest the circle center (Cx, Cy)
t = Dx*(Cx-Ax) + Dy*(Cy-Ay)    

// compute the coordinates of the point E
Ex = t*Dx+Ax
Ey = t*Dy+Ay

// compute the euclidean distance between E and C
LEC = sqrt((Ex-Cx)²+(Ey-Cy)²)

// test if the line intersects the circle
if( LEC < R )
{
    // compute distance from t to circle intersection point
    dt = sqrt( R² - LEC²)

    // compute first intersection point
    Fx = (t-dt)*Dx + Ax
    Fy = (t-dt)*Dy + Ay

    // compute second intersection point
    Gx = (t+dt)*Dx + Ax
    Gy = (t+dt)*Dy + Ay
}

// else test if the line is tangent to circle
else if( LEC == R )
    // tangent point to circle is E

else
    // line doesn't touch circle