Why are we still using audio to communicate between ATC and Pilot?

Question

Amidst the recent Japan Airlines 516 crash at Haneda and multiple others in history like the famous KLM/PanAm at Tenerife, it has always bothered me that these accidents were so easily preventable since they were caused by a simple misunderstanding/communication.

As we know, most of aircraft accidents occur at the landing/takeoff phase.

Why can't we use digital two way radios with text capability which already exist and are very affordable and have been around for a very long time?

UPDATE: There are many answers that seemed to ignore all the notes below because the notes were added later as a clarification to those answers. The most important thing to clarify(where most of the confusion happened) is that this is NOT a replacement, is only an additional communication tool and that this would only be used for specific common instructions and ONLY ATC would do the initial communication(TextToSpeech), pilot would only push a button to ACKnowledge message .

The radio would have a Text to Speech function that would convert the message to audio so that the pilot doesn't have to look at a screen(this also means cristal clear audio, this will also help on the English language mispronunciations/barriers in many countries). The pilot then would push a button on the yoke/stick which will "readback" the message by sending a MSG ID(XXX) ACKNOWLEDGED (each message sent by ATC can be assigned an ID) .

ATC would have a computer with predefined messages(i.e. TAXI VIA A5, HOLD SHORT, CLEARED FOR TAKEOFF, ETC) which would be transmitted to the radio. This will also make the work of the controller faster, more efficient and clear. This will also prevent assigning two aircrafts conflicting instructions since all instructions would be handled trough the computer which will detect if there is a conflict so that the message is not sent.

Besides helping avoid misunderstandings or not understandings, this will also have extra benefits:

• Prevent from talking over.
• Allows to re-read the last instruction (i.e: in case you are talking with your co-pilot) without using the voice frequency to clarify wasting precious talk space specially on congested areas.
• It can also prevent from guessing (i.e. if your talking with your copilot) and ATC sends a message where you think you heard your call sign but it was just another aircraft with a similar call sign)

NOTE: The pilot/ATC would still be able to communicate with voice if they choose to(ie to discuss a special situation) with the same radio they already have.

NOTE2: This is only for communication at the Landing/Takeoff phase which would be within the distance/limit of digital radio transmission.

NOTE3: The pilot doesn't have to look at a screen at all since it will always have TextToSpeech, he/she only has to push a button on the yoke to ACKnowledge the message. If the pilot is unable or wants something else he/she can use the radio with voice normally(no SpeechToText).

NOTE4: For situational awareness purposes, the radio could be configured to TextToSpeech other aircraft instructions that can be cause of a collision . (i.e.: Instructions like Clear for Take Off/land, Cross Runway, Line up and Wait, etc) or it can be configured to Speech all aircrafts instructions if the pilot wishes. (In addition it is worth noting that in theory a conflicting instruction shouldn't happen because the controller will input each instruction into the pc and then converted to audio)

Answer 1

We are not - not exclusively, anyway.

ACARS has been operational for over 40 years by now. No reason to use text-to-speech and reverse, though - if you go text, go text all the way. We have a question on this SE explaining ACARS and related concepts. This article goes into technical details of the protocol and its deficiencies.

Other than converting text to speech, ACARS does exactly what you propose. It communicates in predefined, fixed-format text messages. Every modern airliner has an ACARS terminal, usually located as shown above (A320). Text can be printed out as well, to keep a record.

Voice communication with ATC is a faster, real-time channel* of communication. Two different channels provide dissimilar redundancy - the crew can hear from ATC, as well as read their text messages. Text is better than speech for some things, such as being able to check it again, rather than having to remember. Radio readback is an additional check, where the controller verifies if they've been understood correctly.

Even all two or three safeties fail sometimes. But if there was only one channel of communication, we'd probably have more incidents. This does double the workload, but this is intentional.

ACARS has evolved somewhat over the years, from a simple text screen to a large color display with a QWERTY keyboard:

Using text-to-speech would reduce the overall system's reliability. Humans can only process one speech channel at a time, but we can read and listen simultaneously (to some extent). It's less likely that the same error will occur separately through two distinct channels than through the same one.

Single-button ACK is not practical. Under high workload, simpler tasks get offloaded to Cognitive System 1. System 1 (common to humans and animals) is needed for fast snap reactions, picking the most obvious decision. If you normally press the ACK button 99% of the time, it becomes the automatic reaction in case of cognitive overload. Such reactions have caused multiple crashes.

A verbal readback engages Cognitive System 2, which is verbal thinking, unique to humans, slow and deep. This ensures at least some measure of understanding. To match it with text, one would need to use typeback, but typing incurs a higher cognitive load than speech.

Ultimately, it's a balancing act between cognitive load, speed, and reliability. TTS voice + ACK button could save time and workload, but be very prone to false positives. Text+voice+readback+typeback, separately by PF and PM, would offer maximum reliability, but take the most time (as well as blind-typing literacy and a sidestick). The current system of readbacks plus text information is a compromise that mostly works.

Answer 2

Lets take a few of your points in order

As we know, most of aircraft accidents occur at the landing/takeoff phase and runway incursion is a very common issue.

The first part is correct, the second part is ^{citation needed}. Compared to the number of flights per year, I'd argue that the number is tiny and most of them are minor issues without substantial safety implications. I'm not going to do a full analysis but wiki has a bunch of info on runway incursions. A small excerpt:

Between October 2016 and September 2017, 1,341 [runway incursions] were reported [to the FAA]. Of these, six were placed in the most serious categories A and B. Four of these were considered ATC incidents, and two were "pilot deviations".

The FAA website reports 10,000,000+ movements per year. So 6 serious incidents for over 10 million movements.

Why can't we use digital two way radios

We could, but unfortunately the cost of refitting every radio is prohibitive, and although no doubt beneficial, like everything in business its a cost:benefit analysis.

The radio would have a Text to Speech function ... help on the English language mispronunciations/barriers in many countries

While text to speech has come on in leaps and bounds over the last decade or two. It still struggled with strong accents (obligatory "eleven" video link). So now, instead of misunderstood audio where a controller or pilot can ask for clarification, we have misunderstood text were it could be completely unclear if itts a misunderstanding or a mis-read of Text-to-speech. Add to this the fact that every pilot in the world must be able to use passable English, and they still would in a text-to-speech regime, where is the benefit?

The pilot then would push a button on the yoke/stick which will "readback" the message

This is the part of your question that makes least sense. The point of a readback is not just to acknowledge receipt, it is specifically to read back the receipt so the controller confirms that the pilot has correctly received the message and no miscommunication has occurred. If this becomes automated, there's no way of knowing if the message was sent AND received correctly. I appreciate that with digital communication there are less avenues for misunderstanding, but it still makes no sense to automatically acknowledge a message.

This will also prevent assigning two aircrafts conflicting instructions since all instructions would be handled trough the computer which will detect if there is a conflict so that the message is not sent.

This part makes a lot of sense, but probably aside from all your other suggestions. Albeit the uptake of the aviation industry to new technology is slow due to regulatory demands, I think in the next decade or so we will see AI being introduced to ATC functions, and probably for the purpose of error detection.

Answer 3

In Australia we have the ability to communicate in text form with pilots using CPDLC (Controller Pilot Data Link Communications) and send Pre-Departure Clearances (PDC) via text. I know this is available around the world and not what you have stated but this will reenforce what @randomhead has been saying about the issues with text communications over verbal communications.

So, I am going to answer this in parts based on your statements in the questions.

The Benefits

Besides helping avoid misunderstandings or not understandings, this will also have extra benefits:

• Prevent from talking over.
• Allows to re-read the last instruction (i.e: in case you are talking with your co-pilot) without using the voice frequency to clarify wasting precious talk space specially on congested areas.
• It can also prevent from guessing (i.e. if your talking with your copilot) and ATC sends a message where you think you heard your call sign but it was just another aircraft with a similar call sign.

Yes, this can reduce the risks of helping avoid misunderstandings.

But the talking over each other is replaced with overload on the ATC with the ability for all aircraft to make requests at the same time. I have experienced this with CPDLC, CPDLC sends messages by satellite between the aircraft and the ATC. With each request coming in there is more workload on the controller. You might say the workload always existed but that is not the case.

When you can't hear the other pilots you, pilots can make requests that they may not have made on a busy frequency. Now the ATC has to process that request individually.

Additionally, when you are in congested airspace, I expect you to be paying attention as the instructions that are given are important to be enacted promptly. Most pilots I know have a higher attention in those times to get their instructions. Additionally, I don't mind the occasional Say Again.

The final benefit can be all about callsign confusion. This is can also happen on the ATC side with instructions being sent to the wrong aircraft. So, your replacing one issue which is occasionally a problem which we have procedures to reduce the risk. And replace with a new risk that can happen at any time even when there is no callsign confusion.

The Notes

Note

NOTE: The pilot/ATC would still be able to communicate with voice if they choose to(ie to discuss a special situation) with the same radio they already have.

This is still the case where there is ability to do text back and forth right now and is chosen by both ATC and pilots' majority of the time to be the primary form of communication. The only time I have ever chosen to do this via text-based systems is when there is a complex information to issue (e.g. amended route, pre-departure clearance, weather information)

Note 2

NOTE2: This is only for communication at the Landing/Takeoff phase which would be within the distance/limit of digital radio transmission.

That is the time when lots of things are happening and most text communication systems like CPDLC are not used. The ability for quick information exchange is essential. It is more useful during cruise and currently used over when HF communication is needed in the form of CPDLC.  

Note 3

NOTE3: The pilot doesn't have to look at a screen at all since it will always have TextToSpeech, he/she only has to push a button on the yoke to ACKnowledge the message. If the pilot is unable or wants something else he/she can use the radio with voice normally(no SpeechToText).

This would only be safe if implemented identically to CPDLC, this is that by pushing that ACK button the message is translated by the flight computer and the computer flies the new clearance. If a human needs to interact with the aircraft, then there has to be a readback.

Correcting readbacks is the most common thing I have done in the form of fixing errors in my time as an ATC. Internally of the ATC centre we even readback between the controllers that are sitting right next to me on a recorded line so that everyone is sure that both parties have the same information. Humans make mistakes and the readback is critical in catching the first layer of that mistake.

Note 4

NOTE4: For situational awareness purposes, the radio could be configured to TextToSpeech other aircraft instructions that can be cause of a collision . (i.e.: Instructions like Clear for Take Off/land, Cross Runway, Line up and Wait, etc) or it can be configured to Speech all aircrafts instructions if the pilot wishes. (In addition it is worth noting that in theory a conflicting instruction shouldn't happen because the controller will input each instruction into the pc and then converted to audio)

The situational awareness lost if aircraft are not able to hear what each has been a part of the reason for incidents in the past. One of those is the crash between PSA182 and a Cessna both were on approach but on different frequencies though not the main reason for the crash it was a contributing factor. A simple google search can find more.

Comments

These are some of the comments I want to address directly:

From Gabe:

@randomhead I strongly disagree with you. Remember ATC will have a predefined list of common instructions(TURN X, ASCEND X, DESCEND X) they will not write anything they will just click a button on the computer. The pilot won't write or talk anything he/she would just ACK with the push of a button. They will talk (with voice) only if unable/something else. See NOTE 3

The number of different messages for a standard interface exists for CPDLC in the Australian ATC system, this system was populated with all the normal needed messages and came out with over 300 different messages. This was without the different levels built in. This required a multiple stage menu just to find the correct message. To give simple climb instruction in enroute is a minimum of 6 clicks.

1. Select the aircraft
2. Select the communication capability for that aircraft
3. Select the level change section
4. Select the level message to send
5. Select the level to send with the message (or type it in which is a minimum of 4 keystrokes)
6. Click send

This is slow and clunky.

From Gabe:

@randomhead I disagree with you again. As ATC, remember you have to take into consideration that the pilot also needs to read back the message you just transmitted, and also remember that on each message you and the pilot have to add the call sign, which makes the transmission even longer. If you add all that up, I guarantee you text is going to be much faster. Regarding the point example ASCEND 3000, ASCEND 4000, etc, messages like these are going to be more like a template: ASCEND ___ (the controller would just have to make two clicks, 1st Template, 2nd Custom value)

The adding of the callsign verbally is replaced by me having to read your callsign I know this can be attached to your exact aircraft on my screen but I have since the time of your last message been interrupted by four other aircraft with requests. That means I need to get my mental picture updated with that request is for you and not from someone else.

Conculsion

Air Traffic Controller use the radio to control the speed of information they get from pilots to reduce the risk of overload. But they also use it to get quick information out to aircraft, I can give 5 instructions verbally for each data instruction I give.

It will be a long time until verbal instructions are replaced by text systems as the primary form of communication in approach cells or towers. They will expand first in upper airspace in enroute first but away from major terminal areas that have sequencing. CPDLC started as communication for in the ocean as a replacement for HF which is really, really slow at communication. But it is rarely used where there is VHF coverage because it is easier to give a simple instruction via voice.

Answer 4

When I was a student I asked this exact same question. Likely because I was struggling with controlling the aircraft, navigating, and speaking with ATC all at the same time (English is not my first language).

Now that I am used to it, I'd much rather just talk to them than to fumble around with a keyboard to try to send a message. Also, it helps with situational awareness for everyone to hear each other. Now, if we had neuralink or similar technologies, then I'd be interest ;)