Autonomous guided missiles use sensors to steer to and stay on a target.
The basic components of a missile are the sensorhead, steering, motor, and stabilizers.
Various other takes on these components and entirely new components can further increase the effectiveness and reliability of a missile.

Click here[imgur.com] and here[imgur.com] to go to imgur for some gifs. (Steam restricts to 2mb)

If you want to build AGMs yourself, it is highly recommended to get the mods, NoBounds and Block Scaling Tools. They are not needed but make building easier, and the capabilities of missiles greater.

The sensorhead at the front of the missile is where the sensors used for tracking are located.

If the target activates the left sensor, it will press a key to steer the missile to the left,
if the right sensor, the missile will steer to the right.

Steering can simply be a fan in each direction that a sensor is in. When done correctly the missile will turn to whichever direction(s) the sensors detect the target in.

The motor propels the rocket forward, and can be a fan, or a rocket attached with a grabber. Rockets leave a trail of smoke adding to their cool factor, and can also have their fuel run out or explode if wanted. Props are easier to work with and easy to center.

Stabilizers are any blocks that make the missile more stable in a certain direction. Wing panels, overflow fans, and overflow negative power water cannons, help the missile move in a straight line. And if the missile is horizontal, help it to not fall out of the sky. Another stabilizer is the wheel, which ensures anytime the missile tries to tilt, it will not keep on tilting, instead that tilt will get spun around to the other side, largely canceling minor tilting.

RTC/Return-To-Center:
A setting enabled on steering hinges that will automatically make them return to their original position when input ends.
Overflow:
Setting the power of a block past the amount it can properly store, causing special effects, 10^19 for fans, -10^21 for water cannons.
Center of Rotation:
While it would normally be the center of mass that the body rotates around, due to aerodynamic drag and thrust, the body can rotate at a different point.

Different sensorheads will give different benefits.

Fixed Sensorheads are where the sensors are fixed in place.
Pros:

• Harder to distract with debris, flares, chaff
  
• Work well for slow and ground targets

Cons:

• Usually not as good as tracking targets that make sudden movements, whole body has to move in order to track
  
• Usually requires a lot of sensors in order to track well, not many to lock on

Steering Sensorheads utilize steering hinges to move the sensors. This can be done in multiple ways. Steering hinges are most commonly set to RTC or return-to-center, this halves the amount of inputs that would be needed for the missiles and makes them simpler.

When all of the sensors are mounted on top of perpendicular turning steering hinges, they are a Grouped Sensorhead.
Pros:

• Better at tracking than a fixed sensorhead, two hinges move a few sensors
  
• Uses a few less blocks than an independent sensorhead

Cons:

• Not as good as an independent sensorhead
  
• More prone to getting distracted compared to a fixed sensorhead

When all of the sensors are mounted on top of separate steering hinges, they are an Independent Sensorhead.
Hinges can turn inwards or outwards, inwards turning hinges do not get distracted as easily, but have a hard time keeping track of the target. Outwards turning hinges get distracted easily, but have an easy time keeping track of the target.
Outwards turning hinges are more practical as unless you are using logic gates(that turn on when the sensor does not see something), inwards steering hinges must use return to center to track targets, the speed of which is capped to be relatively slow. And inwards turning hinges require better aim to lock on. This section here will be about outwards hinges.
Pros:

• Better at tracking than a grouped sensorhead, each hinge only needs to move a single sensor, and they can turn in different directions

Cons:

• Most prone to getting distracted

By using logic blocks and a sensor on two perpendicular turning hinges, you can make a missile that scans an area and upon seeing an object, turns to it and darts towards it, a Scanning Sensorhead.
Pros:

• Low block count
  
• Very Light

Cons:

• Scanning rate is not high enough to lock onto quick, mobile targets.

Different stabilizers will give different benefits.

Propeller Blocks:
Pros:

• Do not need to compensate for increased drag with more power (not a big deal)
  
• Can reach higher speeds without breaking

Cons:

• Need more for the same effect as overflow

Overflow is using the NoBounds mod to set the power of a block higher than it can properly store. This can give blocks special behavior. In the case of fans and water cannons, it is drag in all directions. The minimum value overflow for fans is 10^19, and for water cannons 10^21.

Overflow Fans:
Pros:

• Less blocks compared to wing panels
  
• Constant drag lowers the inertia of the missile, allowing it to make quicker turns, better tracking

Cons:

• Can break off at very high speeds (nearing mach speeds)

Overflow Negative Power Water Cannons:
Pros:

• Connection point is unbreakable, can go faster than overflow fans
  
• Less blocks compared to wing panels
  
• Constant drag lowers the inertia of the missile, allowing it to make quicker turns, better tracking
  
• Stronger than overflow fans, every fan might as well be replaced with a negative power water cannon, works better than fans and wings past around 800km/h

Cons:

• Heavier than wing panels and overflow fans

Wheels:
Stabilize with rotation but can break

Spinning Block:
Stabilize with rotation but connection point is invincible

Parts must be in specific places in order for the whole machine to work. So you should think about what position would be best for your parts.
As a warning, unless you are making your missile to a high standard, you won't need to be very precise in placing your parts. And a good amount of the stuff here you can ignore otherwise. But, small changes can add up into big ones.

We know the sensorhead should obviously go on the front of the missile but what about everything else?

Fans, Negative Power Water Cannons:
Steering should be far away from the center of rotation. If you placed your steering on the center of rotation, then your missile would not turn towards a target, it would translate to line up with your target. This isn't what you want as the distance the missile would need to move to lock onto the target is many times more compared to simply turning.

Fans, Rockets:
You may think motor placement wouldn't matter as long as it's centered but that changes when you add inertia.
Imagine a missile facing horizontally with two velocity vectors, it's inertia(up arrow) and thrust(side arrow). Combine the vectors together and you get a velocity vector that wants to turn the missile on its COR(diagonal arrow).
The COR can be imagined as the pivot of a lever. Applying force far away from the pivot is easy, applying force close to the pivot is hard. By moving the position of your thrust, you can change where the diagonal arrow exerts force on the lever, and how it can move the sensorhead.
Before we talked about how it was more efficient for the missile to turn rather than translate, the same applies here. The closer the thrust is to the center of rotation the more force will be needed to rotate the sensorhead, that energy is not lost as it turns into translational energy. The further away the thrust is from the center of rotation the less force is needed to rotate the sensorhead, this helps the missile lock on better. And passively steers the missile to aim at the target, imagine a line rotating on a circle pointing towards a point in the center.
Inertiaᵥₑ꜀ₜₒᵣ + Thrustᵥₑ꜀ₜₒᵣ = Movementᵥₑ꜀ₜₒᵣ, (movement of a point) which in turn can rotate the missile

Wing Panels, Overflow Fans, Overflow Negative Power Water Cannons:
Where the stabilizers are matters as well, the further away from the COR the more resistant the missile is to turning, which is good when you don't want it to stray off course. Drag force works as well on a lever.
It is better to have two sets of stabilizers(-wheel), on both sides of the COR. Imagine the points of drag as pins. If you have a single pin, you can rotate a shape around it. If you have the same shape on two pins, it cannot rotate. The drag force isn't infinite so it can still rotate but it's just made harder.
It actually does work like pins as well, if you stack two overflow fans top and bottom on a ballast it cannot rotate anywhere except for where the two pins do not move, that is rotating left and right. This is why the wheel can rotate the missile without having a hard time.

Wheel, Spinning Block:
Position of the stabilizing wheel should ideally be on the COR, but it does not contribute much in most cases. Only if the missile is inherently unstable from overpowered steering or lack of stabilization, the missile will wobble, causing the thrust vector to change slightly. If the wheel were at the back end of the missile then it would have an easier time making the tip spin in larger movements then if it were at the front. Centrifugal force will attempt to pull the heavier head of the missile outwards, more mass in front of the wheel means more centrifugal force, less mass means less centrifugal force.
Using this may help wobble at either end of the missile.

Note: Missiles for air perform better when they rotate to aim, missiles for ground perform better when they translate to aim.

Rotation can multiply the area which a sensor covers. If you spin a sensor in a circle, the area it has scanned is now a circle. A single sensor can scan the area of many.
Combined with a steering hinge which can move up and down, the area that can be scanned is now a cone. And by increasing the turning limit of the hinge, it can become greater than a hemisphere.
Rotation can multiply the effectiveness of blocks, you can use less blocks for more.

Other components can increase the effectiveness of the missile even more.

Continuous Variable Speed:
Created by putting two forward turning hinges on opposite sides of the body activated by the sensors, with fans on them.
The missile will smoothly speed up when it locks onto the target, and speed down when it loses lock, giving the missile more time to lock back onto the target. Good at increasing reliability and power.

Proximity Fuse/Detonation:
A separate sensor(s) that detonates a bomb when activated. It's a waste of opportunity to not be able to explode when you are within range of your enemy but haven't impacted them. Can be made symmetrically if a bomb and grabber are put on top of a 1000x steering hinge. Crossbows and flamethrowers may work as well.

Thrusting:
When a sensor with a certain range is activated, it turns on extra motors and/or turns off overflow if there is any. Making the missile zip towards the target.

Stepped Variable Speed:
While not continuous, it is more robust and allows for higher speeds. With sensor(s) that detect proximity to the target, activating/deactivating fans or negative power water cannons to adapt to the speed of the craft it's chasing.

As long as something performs the same way as whatever you would like to replace then it would work. If you want to replace steering fans with reaction wheels that would work. You could have the missile float midair with balloons.
A reminder that components do not have to be exclusive to one another, they can be combined to share their strengths.
The downsides of sensorhead types can be alleviated enough to be made irrelevant to the reliability of the missile given that the rest of the missile is made well enough, or their downsides compensated for.
Angled sensors on a fixed sensorhead can have the same effectiveness as hinge sensorheads, they are less smooth in tracking, however they have superior durability which makes them better for higher speeds and maneuvers.
Translation has its place in tracking adjustments for missiles, correcting what would be overshooting movements. And making movement to the target more direct and leading.
Please share your feedback, any comments are appreciated.