Sponsored

SamDoe1

Well-Known Member
Joined
Nov 22, 2022
Threads
1
Messages
122
Reaction score
136
Location
Minnesota
Vehicles
Rivian R1T, Ford GT350
Clubs
 
I don't disagree that a locker would be an improvement. However, you aren't going to get the the same behavior without that physical connection.

Conceptually Rivian may be able to switch from torque control to speed control with the motors or something, but I suspect that they have reasons not to do that.
Of course you can do that. The objective is to apply even rotational speed and torque to both motors on the same end of the truck (or all four if the need arises, though this would be rare) such that the traction control doesn't have to do the thinking to find grip.

The disadvantage of this, specific to the Rivian, is that each motor has a finite amount of power and torque available for each wheel. Being that the Rivian is not a lightweight vehicle by any stretch of the imagination, if the amount of power in each motor that has grip isn't enough to get the truck to go...you're stuck. This is where low range is helpful.

At the end of the day, I wouldn't take my R1T anywhere close to the places I took my old Jeep. It's just not made for that and that's ok as it does a whole host of other things better. With a Jeep (or similar) the live axles help to keep the wheels on the ground to prevent a need for lockers to begin with. In all my time off roading I can probably count on one hand the number of times I've used the rear locker and I never used the front.
Sponsored

 

RexRemus

Well-Known Member
First Name
Chuck
Joined
Sep 27, 2021
Threads
30
Messages
609
Reaction score
1,088
Location
Chicago, IL
Vehicles
2023 R1S
Clubs
 
I think people greatly underestimate just how hard it would be to implement a "virtual" locker on a vehicle without the ability to physically connect the axles.

This doesn't negate the validity of the desire to HAVE that option, but I feel like people seem to think it's just as simple as "make wheels spin at same speed unga bunga!!" it's not that simple AT ALL.

While the "speed" matching part can happen pretty easily (just read the sensors and match the speed) what's lost in that is understanding that the amount of power and torque applied to each wheel (when 3 are freely spinning, or in a more realistic scenario, where 3 wheels have constantly varying degrees of highly limited traction) will be VERY different. It takes much less power and torque to spin a completely unloaded, freely spinning tire at 2 RPM (creeping forward) than it does to move the tire that has ALL the traction and is trying to push the weight of a 7k lb vehicle forward at 2 RPM. Sure VISUALLY they will LOOK the same, they will not WORK the same. And then you're asking the vehicle system and sensors to IMMEDIATELY alter that power and torque delivery perfectly as grip/weight changes on any wheel. You don't have a physical connection to automatically "enforce" rotational limits. It has to all be calculated in real-time from force and rotational sensing. This is a VERY hard problem to solve.

TL;DR - Simulating a physical locker on a 4-motor EV is a lot more complicated than just "RPM matching". It's a fiendishly complex problem that requires incredibly precise (and nearly instant) condition monitoring + power application to all wheels. It might not be impossible to do, but it certainly won't be "easy" or "simple".
 

SamDoe1

Well-Known Member
Joined
Nov 22, 2022
Threads
1
Messages
122
Reaction score
136
Location
Minnesota
Vehicles
Rivian R1T, Ford GT350
Clubs
 
Of course it's not easy but I think you did just answer the question of how yourself. Reading wheel speed sensors is easy and applying proportionally more power to the motor that's lagging in order to enforce that speed is the way to do it. It sounds simple, and it is in that it's a normal closed loop control scheme, but implementation is always harder.

I don't think a perfect real time matching is totally necessary, even a good degree of speed control/matching across wheels would be better than sitting there while the computers figure out what wheel does and doesn't have grip before doing the math to apply power accordingly.
 

agame32

Well-Known Member
First Name
Aaron
Joined
Oct 19, 2021
Threads
6
Messages
243
Reaction score
237
Location
CA
Vehicles
R1T, Model Y
Has anyone mentioned that 2-3 cases where the Rivian finally moved appeared to be from the roller tires sliding to the roller edge and catching traction on that edge? If this is actually what happened, it highlights the problem even more as it would mean it wasn’t *only* the traction wheel causing the movement.
 

RexRemus

Well-Known Member
First Name
Chuck
Joined
Sep 27, 2021
Threads
30
Messages
609
Reaction score
1,088
Location
Chicago, IL
Vehicles
2023 R1S
Clubs
 
Of course it's not easy but I think you did just answer the question of how yourself. Reading wheel speed sensors is easy and applying proportionally more power to the motor that's lagging in order to enforce that speed is the way to do it. It sounds simple, and it is in that it's a normal closed loop control scheme, but implementation is always harder.

I don't think a perfect real time matching is totally necessary, even a good degree of speed control/matching across wheels would be better than sitting there while the computers figure out what wheel does and doesn't have grip before doing the math to apply power accordingly.
I think we all understand the "how", but as you say the implementation is vastly harder than it seems. I have/do work in software dev for major SV companies, and I'm sure a lot of Rivian owners cross into the high-tech demographic - I have a saying for this "easy to say, hard to do". It's always "easy" when someone thinks "I have the idea, I told you the idea, so now you can just DO that, right?"

No. The idea isn't the hard part, ever. Just as you said, implementation is the hard part. Sometimes very, very, hard. Self-driving - I have the idea, so the car can navigate itself around, and it'll use cameras and other sensors to look at it's surroundings and then decide what it should do" - easy, right? Clearly not.

I'm not trying to equate the difficulty with virtual lockers to self-driving, but it's a topic people can understand the difference in "saying" vs "doing" well enough because they've seen how hard it is to actually "do". A virtual locker is certainly not as hard as that, but I just think people believe it's far easier than it actually is. Even saying, and knowing "you just use sensor X and it's a closed-loop" doesn't make the practical application and all the possible gremlins and goblins that pop up when getting it wrong, or having a little too much or too little force applied at the wrong time can send the vehicle tumbling end over end, or off the edge of small trail down into a 150ft valley.

Sure, that can happen anyway, but I almost feel it's better to "know" what the vehicle can/can't do and simply avoid such situations out of a healthy sense of self-preservation, than to (possibly falsely) believe you have the exact same capabilities of the vehicle that went ahead of you, and find out all too late that... no, you don't quite have the same capabilities.

It's like Apple for me - Apple more or less blatantly lies about the security of their products, and the majority of Apple users are not actually very tech savvy, so they truly "believe" they are immune to malware. Hence they act without fear or concern of the consequences and often end up realizing, no, you are not immune. At least even "average" Windows users KNOW to be cautious. So regardless of the actual merits between the systems - it's better to be "fearful" or what might happen, and make a choice to avoid it, than to blindly charge in overconfident thinking nothing could possibly go wrong.

Which is a long way of just reiterating that I think it's a hard problem to solve, with high consequences for getting it wrong. Think tank turn - seems easy, it's not, Rivian pulled it to NOT assume that risk, or give users the false confidence they'd be able to just spin wildly in circles somehow magically perfectly safely. This feels very similar. Doesn't mean it (or tank turn) won't ever happen. But I don't think it's far to say it'll be easy and why don't they have it already? how hard can it be? Pretty damn hard actually.
 

Sponsored

AlexF

Well-Known Member
First Name
Alex
Joined
Mar 10, 2022
Threads
0
Messages
65
Reaction score
68
Location
Colorado Springs, CO
Vehicles
Tacoma, BMW
Occupation
Computer Systems Validation Engineer
I'm skeptical that a single motor, regardless of a "virtual" locker or not, has the power to move a 7-8k pound truck when up against an obstacle or on a steep incline. Especially when in the confines of safe operating parameters for the motor, gearbox, electrical system, etc.
 

quartz

Well-Known Member
Joined
Jan 5, 2023
Threads
7
Messages
203
Reaction score
209
Location
Los Angeles, CA
Vehicles
Honda Clarity PHEV
Clubs
 
There are pros and cons to both vectoring and lockers.. at 60Hz it has the potential to modulate power in real time better than any driver, especially in dynamic conditions.. the lockers allow the driver to take advantage of anticipation and feedback (seeing what's ahead and feeling which part is slipping) to disengage those wheels ahead of time. I think the ideal solution is a combo of both, but you gotta appreciate the added complexity, especially on the software side. It may not make financial sense to implement this even if it's feasible.
 

k3g

Well-Known Member
Joined
Mar 2, 2023
Threads
3
Messages
103
Reaction score
242
Location
USA
Vehicles
R1T
Occupation
Robotics
There are pros and cons to both vectoring and lockers.. at 60Hz it has the potential to modulate power in real time better than any driver, especially in dynamic conditions.. the lockers allow the driver to take advantage of anticipation and feedback (seeing what's ahead and feeling which part is slipping) to disengage those wheels ahead of time. I think the ideal solution is a combo of both, but you gotta appreciate the added complexity, especially on the software side. It may not make financial sense to implement this even if it's feasible.
I think when most of us say "we want a virtual locker" what we're really saying is "make the truck a little more capable". I wouldn't be mad if Rivian came back with a smarter rock crawl mode, something that handles completely unloading a wheel in a more predictable manner.
 

RexRemus

Well-Known Member
First Name
Chuck
Joined
Sep 27, 2021
Threads
30
Messages
609
Reaction score
1,088
Location
Chicago, IL
Vehicles
2023 R1S
Clubs
 
I think when most of us say "we want a virtual locker" what we're really saying is "make the truck a little more capable". I wouldn't be mad if Rivian came back with a smarter rock crawl mode, something that handles completely unloading a wheel in a more predictable manner.
I think this is a better request. And a better way to phrase it. Couching it purely in terms of a "locker" makes sense because we equate new things to the things we know, but there's a fundamental difference in the implementation that might not make sense to try and directly or "perfectly" emulate here.

but desiring the vehicle to handle more situations, more capably, when it certainly seems to have the capacity to do so is a very valid ask. And something you would hope Rivian themselves already know and are ALWAYS testing and improving.

I feel like the Rebelle Rally has been an excellent testing ground - where do you think sand mode came from? Hopefully they do more things like that, and collect more real-world data and improve existing modes and add new ones.
 

OrthoBlock

Well-Known Member
Joined
Mar 19, 2021
Threads
7
Messages
117
Reaction score
206
Location
Philly Area
Vehicles
R1T, X5 PHEV
The hardware is there, but the software is not!

@Dirtcom_AJ Out of curiosity, did you guys have a chance to try holding down the brake pedal while applying power on the "gas" pedal (a common two footed driving technique on ICE vehicles when you want to apply torque on wheels, but at the same time want to have direct control over how fast the car is advancing)?

I have not idea how Rivian would react to this type of simultaneous brake application on the rollers. I have played around with it a bit on my own driveway, and my sense is that it does make a difference on how the truck responds -- but don't have a clue if it would make a sense in a real traction limited slow/no-speed situation. Thoughts?
 

Sponsored

zefram47

Well-Known Member
First Name
Aaron
Joined
Feb 6, 2022
Threads
13
Messages
2,019
Reaction score
3,158
Location
Denver, CO
Vehicles
R1T, C6 Corvette GS
Occupation
Software Engineer
Clubs
 
@Dirtcom_AJ Out of curiosity, did you guys have a chance to try holding down the brake pedal while applying power on the "gas" pedal (a common two footed driving technique on ICE vehicles when you want to apply torque on wheels, but at the same time want to have direct control over how fast the car is advancing)?

I have not idea how Rivian would react to this type of simultaneous brake application on the rollers. I have played around with it a bit on my own driveway, and my sense is that it does make a difference on how the truck responds -- but don't have a clue if it would make a sense in a real traction limited slow/no-speed situation. Thoughts?
Actually did this on the rocks yesterday. Two-footing is seemingly essential trying to climb on the rocks, but it's still very uncomfortable. I confirmed @Dirtcom_AJ roller assessment and that applying throttle faster does give you access to the power. But it's going to take awhile to really figure out how to rapidly apply and then back off the power without bouncing around. In a tricky situation it could make things go pear shaped quickly. I really want to get ahold of an engineer at Rivian and see if they understand what the issue is and whether or not they're trying to fix it.
 
OP
OP
Dirtcom_AJ

Dirtcom_AJ

Well-Known Member
First Name
AJ
Joined
Mar 29, 2023
Threads
6
Messages
52
Reaction score
220
Location
Southern California
Vehicles
R1T
Occupation
Automotive Design
@Dirtcom_AJ Out of curiosity, did you guys have a chance to try holding down the brake pedal while applying power on the "gas" pedal (a common two footed driving technique on ICE vehicles when you want to apply torque on wheels, but at the same time want to have direct control over how fast the car is advancing)?
I have not tried two footing in the Rivian enough to determine whether it's useful. There's a couple more things we'd like to try on the rollers and this would definitely be interesting.

Actually did this on the rocks yesterday. Two-footing is seemingly essential trying to climb on the rocks, but it's still very uncomfortable. I confirmed @Dirtcom_AJ roller assessment and that applying throttle faster does give you access to the power. But it's going to take awhile to really figure out how to rapidly apply and then back off the power without bouncing around. In a tricky situation it could make things go pear shaped quickly. I really want to get ahold of an engineer at Rivian and see if they understand what the issue is and whether or not they're trying to fix it.
That's great real world feedback. One essential technique in rock crawling and offroading in general is keeping a slow steady momentum to be able to bump up just enough over the obstacle. The problem with not having a "locker" style system is you can lose momentum as the traction control system figures out where to send torque. In the video, at the 6 minute mark (linked here Video Starts here as 6 min) you can see how I completely lose momentum when I lose traction on 2 tires. The vehicle nearly stops moving. The traction control impressively figures it out, however, if that obstacle was slightly harder or I had to bump up over a small rock I would not have made it due to loss in speed and momentum. In contrast, the Tundra would have crawled over that in a steady pace without losing momentum. (articulation aside and assuming the same traction on 2 wheels).

I have spent more time thinking about this situation and why Rock Crawl Mode sometimes doesn't feel right. I believe the disconnect between pedal pressure and vehicle speed makes the Rivian system more cumbersome. For example, with the Tundra, I barely applied throttle and the vehicle rolled off the rollers in a speed I expected. With the Rivian, I had to press the throttle down over halfway just to get the vehicle to crawl. In a locked ICE vehicle given a certain pedal pressure there is always the same torque applied whether the wheel is on the ground or in the air.

Example; Let's say it takes "X" pedal pressure and "Z" amount of torque to move the vehicle in a given situation. "X" pedal pressure sends "Z/4" amount of torque to each of 4 wheels to move the vehicle. When you lose traction on 2 wheels then you need to apply MORE pedal pressure "X+Y" to send "Z/2" torque to the 2 wheels with traction to move the vehicle at the same speed. Or "X+Y+B" even more pedal pressure to send "Z" torque to one wheel with traction. Is that where the disconnected feel arises?

So the question is, do we as drivers have to adjust to this pedal pressure and speed disassociation with the Rivian? Or is it something that should be refined and programmed out?

My opinion may be biased due to years of driving ICE vehicles, but I feel that pedal pressure and wheel speed should be programmed to be more in sync when vehicle loses traction. Perhaps the Rivian already does that, but right now, it's not enough. Maybe Rivian can refine the traction control enough to the extent that the pedal pressure and speed disassociation is barely noticeable. Maybe at that point resolution and feedback is not fast enough and it is actually easier to implement a "simulated locker" where the wheels are locked at the same speed.

There is a lot of speculation on how hard it is for Rivian to do this. I don't see why it would not be possible considering they have one of the best traction control systems due to their high resolution of control. Higher resolution than Toyota's ATRAC, yet based on this roller test it's not outshining it enough.

Again, I love this vehicle. The frustration only comes from knowing how much better it can be!
 

zefram47

Well-Known Member
First Name
Aaron
Joined
Feb 6, 2022
Threads
13
Messages
2,019
Reaction score
3,158
Location
Denver, CO
Vehicles
R1T, C6 Corvette GS
Occupation
Software Engineer
Clubs
 
There is a lot of speculation on how hard it is for Rivian to do this. I don't see why it would not be possible considering they have one of the best traction control systems due to their high resolution of control. Higher resolution than Toyota's ATRAC, yet based on this roller test it's not outshining it enough.
I'm wondering if part of the issue may be needing a reference for wheel speed vs forward progress. In a rock crawl situation you aren't making enough forward progress for GPS to work, so if you have one or two wheels that aren't spinning, that biases the knowledge of grip from the two that are spinning. At one point I was on two wheels and kinda flying the other two while trying to crawl up onto a rock wall. The one wheel that really had grip was the driver-front as it was on a rock. Two were flying and the pass-rear was on loose dirt and eventually spun up a little. Without romping on the throttle I couldn't get the truck to apply enough power to move, but then it was very abrupt and hard to control. So yes, I think that Rivian's strategy for rock crawl mode should be closer to how an automatic transmission torque converter works. As you add throttle there's an expectation that you incrementally add power/torque and stop adding it when you begin to make forward progress or you spin all the tires. Maybe Rivian is trying to prevent spinning up a wheel with grip leading to snapping an axle or tie rod. Unsure, but the current situation is not good.
Sponsored

 
 




Top