New Type98/99 MBT thread

[Equation]

FYI "hunter/killer" in reference to a tank is a very specific term. It requires an additional independently targetable thermal periscope which can be used to search for a second target while a first target is already being engaged. For example, the M1A1 is not a hunter/killer, while the M1A2 incorporates a commander's independent thermal sight and is considered a hunter/killer.

Automation. The driver doesn't always have to be fully in control while driving. If the computer spots the thermal images of another threat the driver himself can engage the enemy target himself while driving or the driving can automated to certain extent.

 

[Iron Man]

Automation. The driver doesn't always have to be fully in control while driving. If the computer spots the thermal images of another threat the driver himself can engage the enemy target himself while driving or the driving can automated to certain extent.

Maybe you could program something like this, but this clearly involves a tradeoff where the driver has to divert attention away from driving. This option also could not be used in many if not most circumstances.

 

[plawolf]

This is definitely NOT hunter/killer; it's just a slightly faster standard MBT capability. Hunter/killer capability is simultaneous search and attack; the gunner is attacking one target while the commander is tracking another.

That is either just pointless hair splitting, or you simply didn't understand what I wrote.

The time difference between a gunner independently finding and engaging a target, compared to an automated turret and fire control locking on and engaging a target the commander identified is going to be vary between inconsequential and non-existent, especially if the turret is slaved to the commander's sight initially.

By removing the human element and going full auto, the two man crewed tank could probably engage targets faster and with greater accuracy compared to relying on a gunner.

The only situation where having a gunner may yield real advantages is in terms of search time, where the commander could scan right, while the gunner scans left, thereby potentially halving the time it takes to do a 360 scan.

But that's not hunter/killer related, that's pure situational awareness. But even that could be mitigated by better use of technology.

You could quite feasibly have a second, AI controlled scout sensor turret that automatically scans the surroundings, and flags potential hostiles to the commander by sending him an image of the target, who could then one lick to dismiss or confirm the potential hit, thereby transforming the role of the commander from manual operations to oversight and management of automated processes. The exact same kind of thing that has yielded enormous time savings and productivity gains in all other aspects of human life.

 

[Iron Man]

That is either just pointless hair splitting, or you simply didn't understand what I wrote.

The time difference between a gunner independently finding and engaging a target, compared to an automated turret and fire control locking on and engaging a target the commander identified is going to be vary between inconsequential and non-existent, especially if the turret is slaved to the commander's sight initially.

By removing the human element and going full auto, the two man crewed tank could probably engage targets faster and with greater accuracy compared to relying on a gunner.

The only situation where having a gunner may yield real advantages is in terms of search time, where the commander could scan right, while the gunner scans left, thereby potentially halving the time it takes to do a 360 scan.

But that's not hunter/killer related, that's pure situational awareness. But even that could be mitigated by better use of technology.

You could quite feasibly have a second, AI controlled scout sensor turret that automatically scans the surroundings, and flags potential hostiles to the commander by sending him an image of the target, who could then one lick to dismiss or confirm the potential hit, thereby transforming the role of the commander from manual operations to oversight and management of automated processes. The exact same kind of thing that has yielded enormous time savings and productivity gains in all other aspects of human life.

Clearly you don't understand the concept of hunter/killer. That part I highlighted is EXACTLY the point of a hunter/killer capable MBT, not just some nebulous gain in "situational awareness". The commander searches completely independently of the gunner; i.e. the searches are in parallel instead of in series. As the gunner is setting up to engage one tank, the commander is already sighting in on the next tank. This capability only adds some few to several seconds of time compression from one target engagement to the next, but in tank battles this is sometimes all you have separating you from life and death.

Unless the PLA is willing to allow a computer to automatically engage a target without a man in the loop to confirm a target is actually a target, a two man MBT will never be as fast as a hunter/killer MBT. Your theoretical computer could end up sending multiple false targets to the now information-overloaded commander that a human eye would have immediately dismissed as irrelevant clutter. This is exact same reason banks and other institutions use security image-based verification that a 12 year old could discern but a bot hacking program clearly cannot.

 

[plawolf]

Clearly you don't understand the concept of hunter/killer. That part I highlighted is EXACTLY the point of a hunter/killer capable MBT, not just some nebulous gain in "situational awareness". The commander searches completely independently of the gunner; i.e. the searches are in parallel instead of in series. As the gunner is setting up to engage one tank, the commander is already sighting in on the next tank. This capability only adds some few to several seconds of time compression from one target engagement to the next, but in tank battles this is sometimes all you have separating you from life and death.

Unless the PLA is willing to allow a computer to automatically engage a target without a man in the loop to confirm a target is actually a target, a two man MBT will never be as fast as a hunter/killer MBT. Your theoretical computer could end up sending multiple false targets to the now information-overloaded commander that a human eye would have immediately dismissed as irrelevant clutter. This is exact same reason banks and other institutions use security image-based verification that a 12 year old could discern but a bot hacking program clearly cannot.

Well, I am not sure how to make it any clearer, but the highlighted part is precisely what I am suggesting! That once the commander has verified a target is a target, the targeting computer will automatically calculate where to place the shell, as targeting computers already do today, and also pull the trigger without needing additional human input.

I would strongly disagree with the second part of your sentice, that this would remove the man from the loop, commander is still very much in the loop. He would have reviewed and positively IDed the target to approve the kill.

The main difference would be that there isn't a human eye on target until destruction.

That raises issues, such as kill verification, but that has nothing to do with hunter/killer capability, which such a set up will absolutely allow.

However, even that issue could also be addressed by using the driver.

He can still focus on driving, but will either have a MFD, HUD or even HMD, which would allow a small screen direct feed from the gunner sights to be displayed in without interfering with his primary duty of driving the tank.

Looking at that would be just as distracting as people using SatNav displays in their cars, which hundreds of millions of people can manage just fine without any formal training.

In terms of information overload, well, quite the opposite. I would contend such a system would be far easier on the commander compared to the more traditonal role tank commanders perform.

Scanning constantly looking for anything that might be a threat is extremely demanding.

The human mind is best used to make decisions rather than perform simple, repetitive tasks.

Depending on how much they want to invest in such a tank, they could really revolutionise tank warfare.

At the very top end, you can give the tank commander a VR headset, or fighter pilot style HMD.

The sensor turret could scan at several dozen revs per minute, to allow the creation of a composite real time, enhanced visualisation of the battlefield; or you could have an array of sensors to give true real time 360 minotoring.

The AI will highlight any potential threats, which the commander could look at in greater detail by just turning his head. At the same time, this allows him to be able to manually spot any threats the AI might have missed.

Alternatively, you could keep the same setup, but automate the role of the driver, which is already operational on commercial cars.

Assign 180 degree arch to each the crews (may have to sit them back to back to help avoid locational confusion), and effectively have two
tank commanders.

As secondary duties, one man could keep en eye on the gunnery, while the other make sure the AI doesn't drive them over anything they don't want to.

Automation to reducing man power requirements while improving efficiency is an universal trend.

You have warships with fewer and fewer crews, fighters, especially strike fighters and bombers cutting crew members in favour of greater reliance on technology, to removing the pilot altogether.

It just makes no sense other than on cost grounds to resist that trend with tanks.

 

[FORBIN]

Hunter/killer capability is simultaneous search and attack; the gunner is attacking one target while the commander is tracking another.

For give an idea can engage/fired twice as fast targets, first have it Leclerc, Challenger II, Léo 2 A5 and M1A2, 1995 about.

Edit and right now even for T-14 don' t exist a system for replaced the gunner, in more very sophisticated system are not always the best...

 

[FORBIN]

I think a crew of only 2 is so low for maintain a tank especialy for crawlers.

 

[Iron Man]

Well, I am not sure how to make it any clearer, but the highlighted part is precisely what I am suggesting! That once the commander has verified a target is a target, the targeting computer will automatically calculate where to place the shell, as targeting computers already do today, and also pull the trigger without needing additional human input.

I would strongly disagree with the second part of your sentice, that this would remove the man from the loop, commander is still very much in the loop. He would have reviewed and positively IDed the target to approve the kill.

If you keep a man in the loop, then you by definition are prosecuting attacks in series rather than in parallel, which is the entire point of the independent thermal sight. I'm not sure why you're not getting this point. Every target has to be cleared by a human brain before engagement. Whereas with a hunter/killer MBT you have two human brains working independently on evaluating potential targets.

The main difference would be that there isn't a human eye on target until destruction.

That raises issues, such as kill verification, but that has nothing to do with hunter/killer capability, which such a set up will absolutely allow.

However, even that issue could also be addressed by using the driver.

He can still focus on driving, but will either have a MFD, HUD or even HMD, which would allow a small screen direct feed from the gunner sights to be displayed in without interfering with his primary duty of driving the tank.

This has already been suggested by Equation. Your scenario requires bringing in another human brain to make attack decisions, but you're just getting rid of one brain and adding another, except that this other brain now has to also timeshare attacking a target and driving the tank. Like I said, this may be possible under some circumstances, but how often a driver's brain could be 'borrowed' to attack targets in a complex maneuvering and multi-target environment is an open question to which nobody knows the answer.

Looking at that would be just as distracting as people using SatNav displays in their cars, which hundreds of millions of people can manage just fine without any formal training.

In terms of information overload, well, quite the opposite. I would contend such a system would be far easier on the commander compared to the more traditonal role tank commanders perform.

Scanning constantly looking for anything that might be a threat is extremely demanding.

The human mind is best used to make decisions rather than perform simple, repetitive tasks.

Depending on how much they want to invest in such a tank, they could really revolutionise tank warfare.

At the very top end, you can give the tank commander a VR headset, or fighter pilot style HMD.

The sensor turret could scan at several dozen revs per minute, to allow the creation of a composite real time, enhanced visualisation of the battlefield; or you could have an array of sensors to give true real time 360 minotoring.

The AI will highlight any potential threats, which the commander could look at in greater detail by just turning his head. At the same time, this allows him to be able to manually spot any threats the AI might have missed.

Do you know why the M1A1 turret was initially fitted for a commander's thermal sight but didn't have one? Cost. And that was for just ONE thermal sight. And now you want 5 or 6 of them mounted on a tank to provide 360 degree coverage? I have no doubt you could rig up this hypothetical system and provide the hardware and software to give the commander an F-35-style real-time VR HMD. I also have no doubt nobody in their right mind would try to put this kind of system on a tank and mass produce it.

Alternatively, you could keep the same setup, but automate the role of the driver, which is already operational on commercial cars.

Assign 180 degree arch to each the crews (may have to sit them back to back to help avoid locational confusion), and effectively have two
tank commanders.

As secondary duties, one man could keep en eye on the gunnery, while the other make sure the AI doesn't drive them over anything they don't want to.

Yeah it's already operational on commercial cars. And I do believe a person died because of this, and this was on a paved road with relatively straight and well-demarcated lines, and predictable traffic patterns. Add in open country, unclear paths, uneven and unimproved terrain, no lines for a computer to sight in with, the need to maneuver during combat and evasion, and it gets absolutely ridiculous. A commander can tell his driver to "get behind that hill pronto!" and the driver will have a reasonable chance at guessing which hill he means and how to best get there as fast as possible without being shot first. Good luck with an AI trying to interpret and carry out that command and other similar orders. There will be no automation for tank drivers anytime soon.

Automation to reducing man power requirements while improving efficiency is an universal trend.

You have warships with fewer and fewer crews, fighters, especially strike fighters and bombers cutting crew members in favour of greater reliance on technology, to removing the pilot altogether.

It just makes no sense other than on cost grounds to resist that trend with tanks.

Cost grounds is a very valid reason, as I have already pointed out, not only because there are usually many more tanks produced for a military than planes, but also because the unit cost of a tank is much lower than that of a fighter or bomber, making the cost:benefit ratio of an expensive system very high for an MBT. Greater automation is a definitely a universal trend, but automation is not a solution for every problem. If the new Chinese MBT eliminates the gunner, it will not have a hunter/killer capability. It could partially make up for this with greater automation, but as both you and equation have implicitly admitted to this by having to imagine scenarios that parasitize the driver's attention, there are limits to what can be achieved. Hunter/killer capability is not absolutely essential for a successful tank design, but a 2-man MBT will simply not possess this feature.

 

[plawolf]

If you keep a man in the loop, then you by definition are prosecuting attacks in series rather than in parallel, which is the entire point of the independent thermal sight. I'm not sure why you're not getting this point. Every target has to be cleared by a human brain before engagement. Whereas with a hunter/killer MBT you have two human brains working independently on evaluating potential targets.

It's irrelevant how many sights/humans you have involved. Since tanks only have 1 main gun, by definition, enemy tank targets could ONLY be engaged in series.

All hunter/killer does is compress the next-target engagement cycle time by having the commander go hunting for the next target while his gunner engages the current one.

Short of adding an ATGM mount to the commander sights, he will NOT have the ability to engage another tank in parallel to his gunner. Since that is not done on any operational tank, hunter/kill does not allow parallel engagement. It's simply physics.

But these are only single unit comparisons. It is when you take a more realistic scenario with a tank unit that AI gunnery will really come into its own.

With a squadron of tanks, all the AIs from those tanks could network and share battlefield data and co-ordinate their attacks in ways that will simply be beyond the ability of even the most skilled and experienced human crews.

All threats located by the commanders of all the tanks are fed into a single battlegroup threat database, which all the tanks will share. So it won't just be the commander and gunner looking and finding threats, it will be several commanders feeding threat information to each other. Those commanders could also have the sight lines of other commanders graphically superimposed onto their displays, so they can manage gaps, and be easily ability to know where to look if anyone wanted verification from a second pair of eyes or from a different angle.

The AI could then instantly prioritise engagement targets for each tank in the formation based on the relative positions of friendly and foes to ensure there are no 'doubling up' against a single enemy target.

Such a networked battlegroup will be able to identify and engage targets far more efficiently than a conventional 3-4 man crewed tank force of the same size.

Do you know why the M1A1 turret was initially fitted for a commander's thermal sight but didn't have one? Cost. And that was for just ONE thermal sight. And now you want 5 or 6 of them mounted on a tank to provide 360 degree coverage? I have no doubt you could rig up this hypothetical system and provide the hardware and software to give the commander an F-35-style real-time VR HMD. I also have no doubt nobody in their right mind would try to put this kind of system on a tank and mass produce it.

Cost is never considered by itself. It's always measured against the benefits yielded.

I am sure had you asked around in the 80s and 90s, no one would have believed anyone would be paying 100-200m per plane for mass produced fighter jets, yet here we are today.

You also need to remember that advances in technology often brings down the costs of otherwise cost prohibitive cutting edge equipment. There are thermal camera attachments on sale for a few hundred pounds today, and more recent lab advances offers the potential to completely revolutionise the optics world, and bring prices down far more.

Economies of scale matter very much in this regard, and this is where China has enormous advantages compared to others by being the factory of the world, and able to draw on the massive human and capital pool that has allowed it to create. This was a position America enjoyed to a large extent in the 80s and 90s, which was also the time of its greatest relative military lead compared to the rest of the world, which I believe to be no co-incidence.

Just look at the explosion in application of AESA radars right across the board in the Chinese military. It's not because they got more money than they know what they should do with, but because economies of scale has helped to bring unit prices down low enough for massed uptake of the technology.

With VR starting to go mainstream, and China making much of the hardware, expect the unit price to start falling also.

You are also forgetting that the PLA actually has no problem with spending big money on cutting edge tech.

The Type99 might be cheap compared to western standards, but for the PLA, the jump in price going to it compared to what they were used to buying previously was significantly more than what the western armies experienced moving on to the likes of the M1s, Challenger IIs and Leo's etc.

It is a common misconception held on with great determination by western and Chinese military experts and enthusiasts alike that the PLA is 'cheap'. While that might be true in many if not most cases, that is not because the PLA is somehow allergic to expensive cutting-edge kit, quite the opposite in fact - the main reason it was being so 'cheap' in so many areas was so that it could save enough money on spend on the really cutting edge stuff that would have been way out of their price league had they been more profligate with their funding. Things like the HQ9, Type 99 and Z10 just to name a few.

This 'tradition' of allocating a disproportionately large part of the overall budget to invest and equip a small number of elite units actually makes it more, not less likely that the PLA will adopt revolutionary next gen military advances ahead of western armed forces as China gains ground in terms of both technology and economy.

When a western logisitcs officer looks at the costs of something like what I have suggested, he/she is making the decision based on how much it would cost to equip the entire frontline armoured units of the entire army.

OTOH, a PLA logistics officer might only be thinking of equipping a few divisions.

If the benefits justify it, the PLA is actually the one with the track record of being prepared to make the biggest price/cost leaps.

Yeah it's already operational on commercial cars. And I do believe a person died because of this, and this was on a paved road with relatively straight and well-demarcated lines, and predictable traffic patterns. Add in open country, unclear paths, uneven and unimproved terrain, no lines for a computer to sight in with, the need to maneuver during combat and evasion, and it gets absolutely ridiculous. A commander can tell his driver to "get behind that hill pronto!" and the driver will have a reasonable chance at guessing which hill he means and how to best get there as fast as possible without being shot first. Good luck with an AI trying to interpret and carry out that command and other similar orders. There will be no automation for tank drivers anytime soon.

One AI driver related death compared to how many thousands of human driver error caused deaths in the same timeframe?

Please read up on AI development some more. It's NOT navigating roads that are causing AI the most problems, it's dealing with unpredictable human drives that is the main issue.

In open country, with no silly human drives doing stupid things to watch out for, AI should perform better, not worse.

But again, I was never talking about full automation, I was talking about human directed automation.

Just think of it was pretty much an expansion of the fighter fly-by-wire concept.

The key is the most efficient integration of human and computers so they help to compliment each other.

The human is kept in the loop as the ultimate decision maker, while the computers figures out the most efficient and effective ways to execute the decisions the human has made.

The human makes the decision. Once he has done that, he leaves it to the computer to figure out how to acting it.

That was the case with the hunter/killer example - the commander chooses the target while the AI kills it; it is again the case here.

The driver can decide which is the best bit of cover they want to get behind, or the location that would give them the best sight lines and fire lanes, and the computer gets them there in the shortest time.

If the new Chinese MBT eliminates the gunner, it will not have a hunter/killer capability.

Repeating something endlessly does not make it any more valid. And I have already thoroughly disected and dismantled every single aspect of this assert of yours, while all you have done is keep repeating the same claim.

 

[Akkarin]

To approach this debate from another angle:

Iron Man, do you think that capabilities get lost if the crew size gets reduced from 3 to 2 because of automation ? If so: Do you think these capabilites would be lost because automation is not there yet (e.g. its too hard to auto-aim and fire the gun atm) , or do you think that there is a inherent advantage of having a gunner that can not be overcome by automation ?

Per definition, a tank without a gunner can not be hunter-killer. But thats just semantics and doesnt necessarily tell you anything about the capabilites of the tank.