2018 Stop & Go Cruise Demo

:
2018 Madza CX-9 GT
Quick video showing how great the new Stop and Go Cruise Control is in heavy traffic


P.S. My mileage after about 1,500 miles is currently 25.5 mpg!
 
Wow, super cool. That "Heads Up Display" as you called it is amazing!
 
If you come to a complete stop, will it start you again? I feel like I always tap the gas pedal to get it going from 0; just a blip. Maybe I'm just not patient enough to wait for the car in front to get far enough away for my car to start moving.
 
If you come to a complete stop, will it start you again? I feel like I always tap the gas pedal to get it going from 0; just a blip. Maybe I'm just not patient enough to wait for the car in front to get far enough away for my car to start moving.

Yes, unless you've been stopped for too long. Look in the instrument cluster when that happens and see if you see a green Hold mode (he talks about it in the video). If that happens, you have to touch the gas to get going again.

EDIT - see 3:03 in the video. He presses resume to get going again, I hit the gas pedal a tiny bit.
 
I have the same color as yours. Love it. The adaptive cruise control is awesome. I am amazed by the CX9's very short breaking distance.
 
If you come to a complete stop, will it start you again?

If the vehicle remains stopped for more than 3 seconds, no. The Resume Button (left side of steering wheel) will re-activate the function after Start/Stop has been in Hold mode for >=3 seconds.
 
Quick video showing how great the new Stop and Go Cruise Control is in heavy traffic


P.S. My mileage after about 1,500 miles is currently 25.5 mpg!



Nice vid, Hockeye. FWIW, beware of the stop/go function when trailing Motorcycles, however. Sensors remain functional but the narrower profile (surface area) of the motorcycle reduces overall effectiveness. Getting an opportunity to test this is in stop and go traffic is going to be rare. Most motorcyclist these days lane split at the drop of a hat during stop and go traffic. However, recently I was fortunate to get a motorcyclist last weekend who did not lane split in heavy traffic and remained directly behind the vehicle in front of him. Using stop/go, I could see that the CX-9 had difficulty maintaining its typical one vehicle length distance away from the motorcycle directly in front of me.

It executed several stop/go functions with each one having a measurably different final distance behind the motorcycle. Typically, behind wider vehicles the final stop distance is 1 vehicle length. The CX-9 kept executing stops at increasingly closer distances until one time it began accelerating forward while the motorcycle was already stopped less than 1 vehicle length ahead of me. I was forced to take control with manual breaking to prevent colliding with the motorcycle.

That was enough to let me know that I would not be using stop/go with lower profile "vehicles" ahead of me - such as motorcycles. It works flawlessly with other vehicles that fill the lane space ahead. However, it hesitated to slow down multiple times with motorcycles on different occasions in stop and go traffic, confirming for me that it has difficulty with lower profile vehicles. Moral? Keep that right foot closer to the break pedal, just in case.

The Slingshot Effect

Beware, there is also another condition where stop/go has demonstrated an issue. For lack of a better term and because Mazda has not given it a name, I call it the Slingshot Effect. The Slingshot Effect takes place when you are trailing another vehicle using radar cruise control. That vehicle begins to accelerate rapidly away from you, but does so well under the cruise speed you've set as your top cruise speed limit.

Your vehicle begins to accelerate up to your manually established limit. Suddenly, the vehicle you are trailing makes a very close lane change after tailgating the vehicle directly in front of it. The difference in speeds between those two vehicles is significant but not enough to reduce the acceleration of your vehicle now climbing to its manually set limit. This "slingshots" your vehicle forward, as though the vehicle you were trailing pulled you along its path, slinging you directly into the vehicle it just passed while closely tailgating before it made its abrupt lane change. This scenario is not uncommon as someone might be attempting to make a lane change before the gap they see ahead of them closes - this is why they abruptly accelerate and why they end up so close behind (tailgating) the vehicle directly in front of them before their lane change.

For some reason, the sensors don't pick the replacement vehicle very consistently and if the speeds and distances of those two vehicles are just right, the CX-9 continues to accelerate when it should be decelerating behind the slower vehicle now directly ahead of it.

I've noticed this "Slingshot" effect on more than one occasion and have confirmed it in my 2017 and 2018 Signatures. Be very careful when testing/verifying this in your own vehicle. Or, just be aware of this kind of scenario. When using radar cruise with stop/go, I always keep my right foot in position to reply to any circumstance (brake or throttle) immediately.

A word to the wise.
 
Hey CJ4, thanks for your thoughtful reply. I am always careful with motorcycles in front of me (In my younger days I owned and drove motorcycles). And I have noticed sometimes certain vehicles (like low trailers) don’t always respond to the CX-9 Adaptive Cruise as well as regular vehicles do. (And also in the manual it specifically says “During headway control travel, do not set the system for detection of two-wheeled vehicles such as motorcycles and bicycles”. So I’m very careful (But Thanks!)

I’m sure I have experienced the Slingshot effect also – I know that Stop/Go isn’t perfect. I’m thinking that that happens because the Radar is relatively slow in acquiring a new target. I’m thinking it takes at least 3-5 (LONG) seconds to acquire (And visually display) a new target (When the current target suddenly moves away).

When this happens I seem to automatically sense that increased acceleration and tap the brakes to disengage it. BUT this brings up a question: will the Smart City Brake - stop you anyway from hitting the vehicle in front?

I’ve had a few experiences that seem to indicate that it will.

As much as I love the Stop/Go, as others have indicated, it doesn’t drive as smooth as I do. A normal experience for me is I will be going along at about 50mph tracking the vehicle ahead of me. A quarter mile (or more) ahead I see the traffic light change to Red. I often will tap the brake to put MRCC in standby mode and coast a while, gradually slowing (while the vehicle I was tracking keeps racing toward the Red light.). He gets a long way ahead then stops for the light as I slowly approach him. I then want to engage the MRCC again to let it actually stop me (WHY? So that I am can get into “hold” mode again and can automatically accelerate when the light changes). But if I “Resume” to soon, the system will start accelerating toward 50mph because it does NOT immediately re-acquire the target (But takes that 2,3,4 whatever seconds it find the target again)

If you followed all this, then my point is on a few occasions, I have been (for a few seconds – prior to when it locks on to the target) closing rapidly on that stopped vehicle ahead – and just as I manually move my foot to the brake, the BRAKE warning displays in front of me. I assume that that is the “automatic braking system” doing it’s job – I assume that if I do not actually brake the vehicle, it will do it. But I’m not really sure because I am chicken to wait, I go ahead and brake.

Sorry to be so detailed. I’m still not totally sure if that is how the 2 different systems (Advanced Smart City Brake Support & MCRR) work together?

ANYWAY – I’ll stop now. I just returned from a 300 mile trip and learned many more things about LKA (Lane Keep Assist) – I have many questions concerning it that I’d love to discuss – look for my new video soon!
 
Hey CJ4, thanks for your thoughtful reply. I am always careful with motorcycles in front of me (In my younger days I owned and drove motorcycles). And I have noticed sometimes certain vehicles (like low trailers) don’t always respond to the CX-9 Adaptive Cruise as well as regular vehicles do. (And also in the manual it specifically says “During headway control travel, do not set the system for detection of two-wheeled vehicles such as motorcycles and bicycles”. So I’m very careful (But Thanks!)

I’m sure I have experienced the Slingshot effect also – I know that Stop/Go isn’t perfect. I’m thinking that that happens because the Radar is relatively slow in acquiring a new target. I’m thinking it takes at least 3-5 (LONG) seconds to acquire (And visually display) a new target (When the current target suddenly moves away).

When this happens I seem to automatically sense that increased acceleration and tap the brakes to disengage it. BUT this brings up a question: will the Smart City Brake - stop you anyway from hitting the vehicle in front?

I’ve had a few experiences that seem to indicate that it will.

As much as I love the Stop/Go, as others have indicated, it doesn’t drive as smooth as I do. A normal experience for me is I will be going along at about 50mph tracking the vehicle ahead of me. A quarter mile (or more) ahead I see the traffic light change to Red. I often will tap the brake to put MRCC in standby mode and coast a while, gradually slowing (while the vehicle I was tracking keeps racing toward the Red light.). He gets a long way ahead then stops for the light as I slowly approach him. I then want to engage the MRCC again to let it actually stop me (WHY? So that I am can get into “hold” mode again and can automatically accelerate when the light changes). But if I “Resume” to soon, the system will start accelerating toward 50mph because it does NOT immediately re-acquire the target (But takes that 2,3,4 whatever seconds it find the target again)

If you followed all this, then my point is on a few occasions, I have been (for a few seconds – prior to when it locks on to the target) closing rapidly on that stopped vehicle ahead – and just as I manually move my foot to the brake, the BRAKE warning displays in front of me. I assume that that is the “automatic braking system” doing it’s job – I assume that if I do not actually brake the vehicle, it will do it. But I’m not really sure because I am chicken to wait, I go ahead and brake.

Sorry to be so detailed. I’m still not totally sure if that is how the 2 different systems (Advanced Smart City Brake Support & MCRR) work together?

ANYWAY – I’ll stop now. I just returned from a 300 mile trip and learned many more things about LKA (Lane Keep Assist) – I have many questions concerning it that I’d love to discuss – look for my new video soon!



The "radar" is actually a lot faster at detecting objects - a full three seconds would be disastrous. It is probably a lot closer to 0.1 seconds for detection. The Mazda MRCC uses a millimeter wave radar having a conical projection field with a fixed width at the end of its range. Having some familiarity with radar equipment and technology (of a different kind), I'm guessing that the "resolution" of the radar could be a factor with the lower profile objects within its cone of projection. Radar 101, tells us that the flatter and more reflective the surface of the target, the better the return and its resolution. So, motorcycles will definitely lower the return resolution, given their much lower profile - this is in some ways eon of the aspects of "stealth" technology applied in military aviation.

I don't know if Mazda uses the standard spec for the 79GHz radar or not. But, as you say, they offer warnings for using MRCC under at least one of the conditions you mention above. I can tell you with confidence that the primary problem the system has is mostly due to its ability (or lack thereof) to filter out returns (noise) from the much wider and broader flat surfaces presented by the vehicle directly in front of the motorcycle. Therefore, its constantly trying to make adjustments based on two different distance measurements - one from the motorcycle (which is closer) and one from the vehicle directly in front of the motorcycle (which is further away).

So, the adaptive speed control function gets caught in the middle of all that - or bounced around quite a bit. This would seem to cause the wide variation in both speed and distance from the motorcycle that I observed in my testing/eval - right up to the point where the system simply took off and began accelerating to the limited speed I set without full detection of the motorcycle at a much shorter distance from the radar.

New millimeter-wave radar tech and the encoding necessary to apply it is being developed as we type. I expect some of those anomalies such as my "Slingshot Effect" to become a thing of the past, as soon as the separation of returns improves along with their resolution. That is to say that the processors used to do the sensing and signal input to adaptive speed function(s) will at some point be working with a higher grade of input - because the primary resolutions and noise filtering are so much better than they are right now.

Mazda does not get into too much technical information regarding its deployment of MRCC, but I suspect that I'm on the right track in understanding why the "Slingshot" happens in a Mazda CX-9, where it does not happen in the Volvo XC-90, as just one example. Volvo uses a true 79GHz spec having a wider frequency bandwidth of 4 GHz, as opposed to the 77GHz spec with its 0.5 GHz frequency band. This is a major difference and I don't yet know what Mazda has deployed, or whether they are having problems detecting lower profile targets for another reason.
 
Well the actual radar may only take 0.1 sec to detect a vehicle ahead - but the system as a whole takes 2-3 or more seconds before it displays the vehicle icon on the display and is able to "take control" and start braking, or setting the follow distance, or whatever.

I will watch more closely - but what I'm saying is if you are locked onto the vehicle ahead of you - then that vehicle suddenly changes lanes, and there is another vehicle in front of it within "range" - it does not immediately lock onto the newer vehicle in 0.1 second, but more like 2-3 seconds.

Anyway - I am very comfortable now using the Adaptive Cruise - my bigger concern is the LAS (Lane Keep Assist System) - do you have a good understand of it?
 
Last edited:
Well the actual radar may only take 0.1 sec to detect a vehicle ahead - but the system as a whole takes 2-3 or more seconds before it displays the vehicle icon on the display and is able to "take control" and start braking, or setting the follow distance, or whatever.

I will watch more closely - but what I'm saying is if you are locked onto the vehicle ahead of you - then that vehicle suddenly changes lanes, and there is another vehicle in front of it within "range" - it does not immediately lock onto the newer vehicle in 0.1 second, but more like 2-3 seconds.

Anyway - I am very comfortable now using the Adaptive Cruise - my bigger concern is the LAS (Lane Keep Assist System) - do you have a good understand of it?

This is my experience as well.

I think of it as allowing a 16 year old to drive. Watching the cars ahead of me, I see their brake lights and I see them begin to slow down and then 1, 2, 3 and the Mazda system starts to slow down. Braking and acceleration are also about as smooth as a 16 year old...

It works most of the time - but it does not appear to see stationary vehicles sitting in traffic coming up ahead.

The LKAS is mostly useless. It does give you a warning, but it's so late that you're already in the process of crossing over the line. Theoretically, it can intervene and push you back the right way, but other than a small initial push it doesn't do anything. And it only sees the two lanes maybe 60% of the time - granted it's winter in New England, but it wasn't much better before the snow started falling. I leave it off most of the time.
 
Well the actual radar may only take 0.1 sec to detect a vehicle ahead - but the system as a whole takes 2-3 or more seconds before it displays the vehicle icon on the display and is able to "take control" and start braking, or setting the follow distance, or whatever.

Interesting. Had not noticed that icon. Thanks - I'll start checking it out to see if mine does the same. I wonder why I have not noticed it.



I will watch more closely - but what I'm saying is if you are locked onto the vehicle ahead of you - then that vehicle suddenly changes lanes, and there is another vehicle in front of it within "range" - it does not immediately lock onto the newer vehicle in 0.1 second, but more like 2-3 seconds.

The system does not actually "lock on" - but I get what you are saying. Its actually a computational algorithm (in most cases) and its more like sorting, ranking and filtering radar return intensity from a so-called "field" of returns. The system is constantly receiving returns and sorting them out. Those having the shortest return time intervals are given priority - that should be the vehicle you are immediately trailing. When that vehicle moves left or right of the field, its returns are momentarily no different than any other vehicle directly in front of you. Once its either moves far enough laterally outside the field (which is conical), or far enough out in front of your vehicle such that its returns are now longer than those of other objects, its returns cease being the priority. This does not however, address the issue of RCS or Target Size and its impact on decisions ultimately made by the MRCC about throttle position.

Only one of the keys here is the millimeter wave frequency band that Mazda has selected for deployment into their vehicles. This discussion can get very technical (if you like), but suffice it to say that in most automotive applications that I am aware of, there are essentially three (3) options: 24/77/79 GHz. 24 GHz MWF is limited to only 200 MHz bandwidth. 24 GHz is still used for other perimeter type functions such as Lane Departure Assist. However, for collision avoidance, 77 GHz was used and now 79 GHz is ultimately going to replace it given its wider bandwidth and its ability to probe further out ahead of both 24 GHz and 77 Ghz bandwidths.

Bottom line, think of 79 GHz in the automotive world as you would 4K video and the other three as you might 1080p and 720p. Those are not exact equivalents, but it sets the stage for thinking about Automotive Radar in terms of its ability to provide "Clarity" and "Resolution" of objects being detected - which then affords the ability of engineers to develop more capable Decision Support Algorithms that go into systems such as the Mazda MRCC, as just one example.

So, this should help to understand where that 2-3 second perceived delay might come from at least in part. There are other reasons that contribute to the perceived delay, however. I say "perceived" because the Radar itself is not failing to detect. It could be doing a fine job of detection, where the decision support system behind it that makes throttle position determinations is not yet "satisfied" with that the Radar returns and thus does not make a throttle or braking decision in-line with what you might have done as a driver yourself. These systems will improve over time. They will get smarter, sharper, clearer and more capable of simulating simple decisions a human might make under a given set of road based circumstances.

By the way, if you have not demo'd the Volvo City Safe system, I would highly encourage it just for kicks. It is quite different than Mazda's MRCC in function and purpose. It also gives you a good idea of what's ahead in auto safety features that will eventually become standard equipment and/or even regulatory requirements in the not so distant future (I would venture to guess).





Anyway - I am very comfortable now using the Adaptive Cruise - my bigger concern is the LAS (Lane Keep Assist System) - do you have a good understand of it?[/QUOTE]
 
It works most of the time - but it does not appear to see stationary vehicles sitting in traffic coming up ahead.

Yep, I've noticed that too. Rather startling the firs time it happens. I closely monitored several additional self-induced evals of this 'failure' and subsequently changed my use of MRCC as a direct result. Thus far, I have found that MRCC/S&G works (for me) fine on the open road without any issues (day or night). It works great on the freeway in heavy traffic as long as I stay on top of the Resume Button. However, once traffic begins making close lane changes in front of me, that's when Hockeye's statement (at least in my driving experience) becomes mostly true.

The two biggest problems I see are the "Slingshoot" and the outright miss (that you just mentioned) of a fully stopped vehicle in the distance while MRCC is accelerating up to the limit speed you set. That's one of the scariest and I no longer allow the system to have control when traffic conditions present such a scenario. If I see a large gap developing like that with a vehicle stopped in the distance while in heavy traffic, I just pause the system, take full control and then reengage the system once traffic has 'smoothed' out a bit and become more 'predictable' for the system.


The LKAS is mostly useless. It does give you a warning, but it's so late that you're already in the process of crossing over the line. Theoretically, it can intervene and push you back the right way, but other than a small initial push it doesn't do anything. And it only sees the two lanes maybe 60% of the time - granted it's winter in New England, but it wasn't much better before the snow started falling. I leave it off most of the time.

Hmmmm. Interesting. LKAS has been outstanding in my 2017 and 2018. We've had no such snow on the ground around these parts, however. Makes me wonder if the sensors are having an issue with the snow somehow. I leave mine on because it helps so well with blind spots - shoulder checks and LKAS are a nice tandem around here. We have really bad drivers who tailgate and then make snappy lane changes as if they are drafting at Daytona (really dumb). And, it has only gotten worse over the years around these parts.
 
Back