“Crawl Control” for the Cybertruck?

Zabhawkin

Well-known member
Joined
Sep 1, 2021
Threads
4
Messages
163
Reaction score
237
Location
New Mexico
Vehicles
1999 Nissan Frontier, 2015 F-150, 1984 Jeep CJ7
Country flag
I want full digital remote for rock crawling. So I can get out and see exactly where things are and make fine controls while outside of the vehicle and can see the hazards I’m going over.
Via phone app, left thumb pad front throttle and steering, right thumb pad rear throttle and steering limited speed to 2mph.

Edit to add: This would be awesome for backing trailers into tight spaces as well.
 


slomobile

Well-known member
First Name
Dustin
Joined
Apr 7, 2022
Threads
0
Messages
48
Reaction score
63
Location
Memphis
Vehicles
Cybertruck
Occupation
Roboticist
I want full digital remote for rock crawling. So I can get out and see exactly where things are and make fine controls while outside of the vehicle and can see the hazards I’m going over.
Yes! But why get out? Launch a drone from the vault to take a peek while you watch the screen in comfort.
 

slomobile

Well-known member
First Name
Dustin
Joined
Apr 7, 2022
Threads
0
Messages
48
Reaction score
63
Location
Memphis
Vehicles
Cybertruck
Occupation
Roboticist
JBee, I'm trying to understand the system you describe because it would make my projects easier if true. Is this something that already exists, or theory? Can you steer me toward some reference material?

optimizing for each wheel individually to the particular piece of ground it is traversing at that time is necessary to achieve optimum performance.
Its like the movie A Beautiful Mind where in the bar, John Nash says Smith needs revision. Yes, optimize each wheel individually, but only so far as it advances the goal of the entire vehicle/driver system.
If left wheels have low traction and right wheels have good traction, you can optimize each wheel for traction, but that results in driving left hand circles only. Once you overlay steering and throttle, wheel traction is forced to be non optimal. At best, you can identify the max non slipping RPM of the worst wheel, and scale down the rpm of all other wheels to match. This results in something like ABS. It is disturbing as a driver when it engages. Suddenly your accelerator seems like it isn't working. To test it, you floor it, and the vehicle doesn't go any faster. Then the bad wheel exits the frozen puddle it was on, and the whole system no longer limited, rockets forward into a tree.
So depending on the traction coeffcient of each wheel at that particular time, torque can be applied, and a slip ratio can be calculated to understand the traction available on each surface under each wheel. Once you have that you simply overlay the steering input to modulate the radius difference each wheel has to travel when turning, to yaw the vehicle in the direction you want, up to the physical limits the vehicle can calculate too.

Essentially, an EV with a motor per wheel, and air suspension (with load sensing, which most have) it can calculate exactly how much traction is available per wheel and command the motor controller to create the right amount of torque at the desired RPM for each wheel.
How can the vehicle know "the traction coeffcient of each wheel at that particular time"? It can estimate if it already knows the coefficient of friction of the road surface. But that is the trick. Making a bad friction estimate spoils all other calculations. Pavement is fairly uniform so an average value can be used, off road crawling has highly variable friction coefficients.
When I've measured coefficient of friction for tires against various surfaces, it involves dragging the loaded tire across the surface with an inline force sensor and noting the force applied at the moment static friction drops to become sliding friction. This measurement process could easily be adapted to note current required to spin a tire, but I hardly think you want to drive a vehicle that intermittently spins its tires to sample the available coefficient of friction. If there is some other way to measure that, please share. Now I want to have a "road surface coefficient of friction measuring mode" as well.

I am certain the CT will not have independent non-linked front or rear steering, so each wheel can go it's own way, simply because it will need a akerman linkage to join either side to do low scrub turns and would otherwise not be road worthy.
I'm confident you are correct about that. But it is the nature of some trucks to be modified by owners to suit strange purposes. I am definitely one of those idiots that will chop it up and turn it into something ridiculous.

Tractor crab steer for less soil compaction was a great example. With cameras in the control system it could become more intuitive to have one wheel hug left of the row and other wheel hug right of row. Depending on row spacing, you could have each wheel in a different row.

With cameras watching the border between plants and aisle, manual steering authority could be restricted to within the aisle unless brake is pushed to override the mode and allow driving over rows to avoid obstacles. Crabbing might also feel more intuitive if the cab swiveled to always point in line with the wheels. Then it would appear as if you turn the wheel right in order to spin the constellation of wheels to the left. But thats just theory, I'll defer to your experience with 4ws on big equipment. I'm envious of the research opportunity. What do you suppose the reaction would be If I asked a random local cotton farmer for a ridealong?

End of row turning areas are ripe for optimization.
Not CT, but same self driving tech generally.
 
Last edited:

Ogre

Well-known member
First Name
Dennis
Joined
Jul 3, 2021
Threads
107
Messages
6,557
Reaction score
15,455
Location
Ogregon
Vehicles
Model Y
Country flag
Yes! But why get out? Launch a drone from the vault to take a peek while you watch the screen in comfort.
Six of one half a dozen of the other. Phone remote would have the advantage of not needing any additional hardware. Maybe you could put it in a remote mode and use a paired Bluetooth game controller.
 


rr6013

Well-known member
First Name
Rex
Joined
Apr 22, 2020
Threads
43
Messages
1,249
Reaction score
1,205
Location
Coronado Bay Panama
Website
shorttakes.substack.com
Vehicles
1997 Tahoe 2 door 4x4
Occupation
Retired software developer and heavy commercial design builder
Country flag
I've written thousands of lines of code doing this and the hardest part has been the user interface. Frankly I'm pretty bad at it. There are just too many options. Simplifying the user interface by limiting your options is one very effective way. It is not the only way, but may be best for the masses.
SteveJobs is instructive having spent a decade in unix with Berkeley Std Distr many, many options. SteveJobs created NeXTstep UI to simplify those unix options. Still the base level knowledge exceeded human. NeXTSTEP UX was automated eliminating the “human” requirement merely taking orders and carrying out the desired end result. At the end of a decade SteveJobs resigned himself to the fact that in a mass market people are happy with simple flavors even if that’s only two choices. There isn’t space or way for a third even if it is better, cheaper.

Hence, returning to Apple SteveJobs castrated unix, assassinated API’s and reduced the supportable functionality down to human scale. That exactly opposite of the situation that you have presented programmatically.

Tesla can’t support programmatic executables over electronics. As a research project? Driven by problem-defined improvements in code that human control can choose - that’s a grant candidate.
Driven by automation improvements in code that eliminate human input -that‘s a secondary side effect of the first goal improving a performative quality. Maybe the programmatic is not needed. So that’s question #1.

Cyber in Cybertruck is MacOS-like simplifying the UX, extending its capabilities and networking the platform in ways that do work otherwise performed manually. Cybertruck subtexts Armageddon. Performance, survival, off-grid and tougher than the average truck. Picking MOAB is admirable, a challenge and fun model at scale to play with.

Bottomline: Does it sell Cybertrucks? Does it create a Tesla brand of trail-rated off road safety? Can it self-recovery without human input?

Those are tricks Elon Musk can buy into.
 

slomobile

Well-known member
First Name
Dustin
Joined
Apr 7, 2022
Threads
0
Messages
48
Reaction score
63
Location
Memphis
Vehicles
Cybertruck
Occupation
Roboticist
Bottomline: Does it sell Cybertrucks? Does it create a Tesla brand of trail-rated off road safety? Can it self-recovery without human input?

Those are tricks Elon Musk can buy into.
I am not suggesting that scriptability would be, or should be used by the masses. I am suggesting that a small number of very dedicated Tesla enthusiasts can contribute value to the brand by independently developing new use cases for CT by way of the scripting API.

It is akin to factory support of racing endeavors. Bench racing perhaps in this case.
The scripts developed would be invaluable to those that actually need them, because there is nothing else like it. That may not have much real value to the average consumer, but it has a great deal of perceived value for the brand. With Tesla's current image of being unfriendly toward independent, or self repair, an open API could counter that to some degree. It has intangible, though profitable benefits. Same with racing.

If you need a computing example, I'll stipulate that Apple has the best mass market consumer interface. But when a dedicated enthusiast needs custom software for a one off use case, they are NOT turning to Apple. They are building it on a Linux. If what they build is good, Apple will port and polish it eventually and sell it as their own. Garage band...
 

Ogre

Well-known member
First Name
Dennis
Joined
Jul 3, 2021
Threads
107
Messages
6,557
Reaction score
15,455
Location
Ogregon
Vehicles
Model Y
Country flag
Tesla cares first and foremost about scaling.

Enabling hacking the truck is not something that they are going to invest in. It would be cool, but I suspect controlling your truck via a blue tooth connected game controller is more likely.
 

rr6013

Well-known member
First Name
Rex
Joined
Apr 22, 2020
Threads
43
Messages
1,249
Reaction score
1,205
Location
Coronado Bay Panama
Website
shorttakes.substack.com
Vehicles
1997 Tahoe 2 door 4x4
Occupation
Retired software developer and heavy commercial design builder
Country flag
Six of one half a dozen of the other. Phone remote would have the advantage of not needing any additional hardware. Maybe you could put it in a remote mode and use a paired Bluetooth game controller.
At most, Cybertruck will FOB PUREvision for “Follow Me” mode. Tesla will not surrender to iPhone walking a Cybertruck, ‘cuz optics. And FOB Follow Me sure makes going thru locked gates a piece of cake, feeding hay to critters in winter and beats walking back after dropping a tractor off at the neighbors’ cross the road.
 

Zabhawkin

Well-known member
Joined
Sep 1, 2021
Threads
4
Messages
163
Reaction score
237
Location
New Mexico
Vehicles
1999 Nissan Frontier, 2015 F-150, 1984 Jeep CJ7
Country flag
At most, Cybertruck will FOB PUREvision for “Follow Me” mode. Tesla will not surrender to iPhone walking a Cybertruck, ‘cuz optics. And FOB Follow Me sure makes going thru locked gates a piece of cake, feeding hay to critters in winter and beats walking back after dropping a tractor off at the neighbors’ cross the road.
I was thinking something like a while back calling it puppy mode.

 

 
CYBERBACKPACK
Top