Seems like they should be a transceiver + retroreflector, not transmitter + receiver pair. Is there a cost or safety reason for this?
A reflector might accidentally be substituted by some reflective aspect of an obstacle, say a chrome plated fender on a bicycle, motorcycle, classic car... or even the reflector on one.
In contrast a transmit receiver pair is far more likely to indicate the true and complete path.
The typical LiftMaster/Chamberlain safety sensor system is designed to prevent it being accidentally or even easily deliberately disabled.
Interposing a reflector, an electrical short or an electrical open anywhere external to the controller will not provide a false 'clear' signal. In fact no combination of passive components connected to the sensor terminals can defeat the safety device.
The main reason, for using a through-beam sensor for a garage door, would not be cost but safety, reliability and trouble-free operation.
A retro-reflective sensor could be considered unsafe for this application, should one take into account the rare event of it responding to the reflection from an obstacle.
For a through-beam sensor, the actual distance traversed by the beam would be its range, whereas, in the case of a retro-reflective type, the actual distance would be twice the range. Secondly, for the retro-reflective type, loss of signal in the reflector would also need to be factored in. Hence the typical range of a through-beam sensor would be more than twice that of a retro-reflective one.
Thus, compared to a retro-reflective sensor, the through-beam sensor with its higher excess gain (more transmitted light, than that required to activate the receiver, impinging on it), would give consistent and trouble-free operation even in dusty and dirty environments.
Thus, inherent safety and higher reliability of the through-beam sensor would dictate its usage, notwithstanding its higher cost.
Instead of a retroreflective prism or tape, a set of mirrors could return the beam to the combined transmitter and receiver that receives a beam from above. That would overcome the "safety" objections posted by Luddites above. Three simple flat mirrors at 45 degrees is all it takes-- no wiring. Of course my trailer hitch would still interrupt the beam and prevent the door from closing and the additional mirrors would be additional spider web accumulators.
This may be a great idea for product improvement!
There are some very simple chips out there to ease the design of a light beam, that bring the cost down and ease of use and reliability up - it's a solution that's been in use for more than 4 decades at least. So nothing new or exciting about that.
However, using a retro-reflector requires sending out a very very short light pulse (assuming a 5m distance, the reflected pulse, now greatly attenuated by distance, will return in ~33 nanoseconds, requiring measurement of nanosecond pulses with phenomenal accuracy (ideally you'd need to detect anything more than a 10cm variance? How big is a leg in white trousers, or a hand with reflective glove? 10cm corresponds to something like 300 pico-seconds!). At install, the sensor must be calibrated or "trained" to open a detection window of a few 100 pico-seconds at the precise time to expect the response. Some kind of "learning" algorithm is required. Also, the returned light needs to be distinguished from ambient light clutter and noise.
This can all be done with modern electronics and optics, however, it would require a fast micro-controller, some high-spec photo-diodes and perhaps even a laser LED capable of generating extremely short pulses. 20 years ago I used a laser distance sensor that used a multi-watt laser (and a big lens), pulsed for an EXTREMELY short time, and then just a PIC micro-controller that used a huge amount of oversampling to improve the measurement accuracy. Technology has certainly moved on since then.
Why would you use a retro-reflector? Because, despite being more costly to make, it is potentially easier to install (one less set of wires to run). In my experience install costs usually dwarf the cost of the electronics for most household installs - in most 1st-world countries labor is expensive.
Plus, the installer often chooses the brand of product to install - most installers will favor the more expensive unit that is easier to install. So this may be an idea whose time has come - well done Tom for asking this question.