Prius: Do not disengage openpilot when pressing the gas pedal #393
Comments
|
This is part of openpilot safety requirements:
See https://github.com/commaai/openpilot/blob/devel/SAFETY.md |
|
I would agree with mtourangeau. OEM toyota system do not disengaged DRCC by pressing gas pedal. it will be so helpful in stop and go traffic. without pressing Resume button all the time. |
|
How does pressing the gas benefit you in stopped traffic?!? Is it because of the minimum engage speed? When your car has real stop and go, like when you add Comma Pedal, you literally never disengage because you never need gas or brake the entire drive.
I think that any code to change gas disengage is not helpful to the end game and just creates further safety issues.
Yes it’s annoying when you don’t have actual low speed follow / stop and go but get Pedal and you’ll have it. :)
Just my two cents.
…On Oct 18, 2018, 2:02 AM -0400, fkong777 ***@***.***>, wrote:
I would agree with mtourangeau. OEM toyota system do not disengaged DRCC by pressing gas pedal. it will be so helpful in stop and go traffic. without pressing Resume button all the time.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
I just think it seems like more of a paradigm shift that you’re having to get accustomed to than it is anything else…
After pressing the gas, OP disengages since you’re no longer self-driving you’re asking to take over.
At that point, simply press the button again to continue your journey… its as simple as that.
Also, OP doesn’t fully engage until you let off the gas after engaging OP.
TLDR; why is it so hard to just press the silly button again to set your new speed to the speed you accelerated to?
Seems like a silly problem to “fix” considering the massive disadvantages it causes from feeding bad data about acceleration.
This is the Tesla model which is pretty much I think the way the future of autonomous driving looks like.
You’re either in “openpilot mode” or you’re not. One or the other.
The only reason steering doesn’t disengage is because it is useful for Machine Learning and fixing mistakes.
Staying engaged while accelerating messes with the data I’d imagine and is no longer self-driving.
We’re just having a discussion… its a discussion that is worth having as others have also complained about this.
But again, my two cents are that its just a matter of getting used to driving with OP versus with the Stock systems.
The stock systems are bandaids put in place by the car companies for various reasons but their intentions aren’t the same as OP.
OP appears to have more of the Tesla model which is the end game is, we shouldn’t need to touch anything. Where as, the stock system that so many seem to love requires you to basically be tethered to the steering wheel AS MUCH AS POSSIBLE so that your family doesn’t sue the car company when you inevitably die from not driving your own car with the shitty stock system that is presently in the majority of our cars.
It’s a trade-off for sure with Openpilot… but I think it’s the way to go long term. “BIG PICTURE” type of view.
Healthy discussions. But in the end, Comma needs the data in a clean manner to build accurate maps and planning code.
Without that data, OP doesn’t exist and stays the same old LKAS/ACC features you get from your stock system.
THE ENTIRE PURPOSE of using Openpilot versus your stock system is OTA Updates, Better Vision Model, Longer engagements without having to do anything.
It’s NOT its purpose to just replicate your stock features all together.
And I’m eternally grateful for that, as the majority of stock systems put in cars are currently extremely limiting and will be for many years to come since most car companies aren’t yet willing to roll out something like what Tesla has.
<3
…On Oct 18, 2018, 8:47 AM -0400, Mathieu Tourangeau ***@***.***>, wrote:
Basically when the traffic stop completely the car stop. Then you have to press the gas pedal to continue or press the resume button. It would be nice if open pilot can continue to control or resume by it self.
Also, when you in fast traffic like 45-55MPH, and you want to by’ass one one or what ever then you use the gas pedal and open pilot stop.
It’s a useless discussion since OP will never do it. I’m going to delete this issue.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Also, should be noted that much of what I just said is based on my limited understanding of Openpilot and Comma’s overall intentions from what I’ve seen and experienced.
It is in no way an official response or meant to be definitive. Just talking here folks.
…On Oct 18, 2018, 8:47 AM -0400, Mathieu Tourangeau ***@***.***>, wrote:
Basically when the traffic stop completely the car stop. Then you have to press the gas pedal to continue or press the resume button. It would be nice if open pilot can continue to control or resume by it self.
Also, when you in fast traffic like 45-55MPH, and you want to by’ass one one or what ever then you use the gas pedal and open pilot stop.
It’s a useless discussion since OP will never do it. I’m going to delete this issue.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Yikes okay… I think I’ve wasted my time.
<3 cheers mate.
…On Oct 18, 2018, 10:18 AM -0400, Mathieu Tourangeau ***@***.***>, wrote:
I trust more my stock breaking system then open pilot breaking system. For me, OP drive and should’t care about breaking and accelerating since that the stock system that do this functinality.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Closing this issue. |
|
@mtourangeau @fkong777 see this commit for the possibility of stop and go on Toyota without user input, it works on my Prime (be sure to read comments): https://github.com/zorrobyte/openpilot/commit/4b70e41e9b87c6a01f7fff3f018ab1739e0038f2 |
emmertex
pushed a commit
to emmertex/openpilot
that referenced
this issue
Jun 22, 2019
add a_rel to data logger
|
@zorrobyte the commit works, but it will use a LOT of battery as the car never truly "stops", just give the cruise lever a wiggle instead |
pd0wm
pushed a commit
that referenced
this issue
Dec 14, 2019
a648cca Add os import 042562d Extracted wifi connect from test helpers ac0fd5d query fw versions example - use extended diagnostic session 4e9d788 Remove not-needed cadillac-init f0a5d15 typo c093286 Add bootkick after re-enabling phone power (#401) eadb0db security upgrades (#397) 7c13bec Command to get signature (#399) dad439a static assert on size of health packet (#398) da9da46 Fix VERSION df4159c Revert "Revert "Register readback on most modules. Still need to convert the other ones (#396)"" 56ec215 Revert "Register readback on most modules. Still need to convert the other ones (#396)" 893e486 Register readback on most modules. Still need to convert the other ones (#396) 6bbae7b VW safety: allow cancel spam on both buses to be compatible with camera and gateway integration d5f7a28 bump panda 1bcc351 ignition_can: set it to False after 2s of not seeing CAN msgs 96137f1 VW can based ignition not needed. it has ignition line at camera as well. 1b004a1 Same flake8 version as the one in openpilot e82ba5f Same pylint version as the one in openpilot 656f99b Interrupt refactor (NVIC_SM_1: #334) and Fault handling (#377) (PR #373) 000282e Fix can_logger.py to run correctly on python3 (#392) 7f9b4a5 Fix USB device enumeration on Windows 8.1 and Windows 10 (#393) dec565c Update Misra test coverage, which now includes rule 2.7 fb6bc3b Fix Misra 878dd00 solve race condition is relay_malfunction right after changing the relay status by adding a counter 2d4cb05 add a safety mode counter a6797a21 Implement USB power mode on uno 670f90c Merge branch 'master' of github.com:commaai/panda ca39a5d Added faults integer to health packet e1c34a1 Panda Jungle testing (#394) 2a093a3 Added heartbeat to echo test 2246435 Fixed health struct size. We should really get an automated test for this f458d67 Add uptime counter to the health packet (#391) 1662481 enable CAN transcievers outside the set_safety_mode function, which is not related a7c9874 bump panda ver 1192d93 Power saving refactor (#389) d58d08f Fix Misra 17.8: can't mod function params bc685ac Minor indent a54b86c Failure of set_safety_mode falls back to SILENT. Failure to set silent results in hanging 597436d NOOUTPUT safety mode is now SILENT. NOOUTPUT still exists but keeps C… (#388) d229f8d ESP forced off in EON build. this prevents ESP to be turned on when e… (#387) 8a044b3 forgot Hyundai: now also using make_msg 4f9c879 remove abunch of lines from safety regression tests by using common make_msg function fb81414 mispelled word 57f5ef8 Fix misra: addr can't be more than 29 bits anyway 68ff501 typo d5c772b Fixe Toyota message white-list 48197a9 Better masking for ELM mode b8fe78c VW is also tested for safety replay 212d336 Safety Chrysler: Added cancel spam button check d44b562 fix print in example 02d579a functional addr handling 6249a18 tx_hook shall have a white-list of messages (#381) 8138fc1 uds: handle function addrs and fw version query example 6626a54 Fixed python health api b9b79e8 uds zero second timeout e0de1a4 define ALLOW_DEBUG in safety tests 86dec4b Safety modes that violate ISO26262 requirements are not compiled in RELEASE build e74ed93 safety tests a bit more simplified 2027765 relay malfunction test centralized 8af1a01 clean up safety tests e8f7a3b upd panda cfcce8f WIP: Relay malfunction (#384) 69d9d61 No tabs in mazda safety a86418c insignificant changes f239b99 single addr was better d063a18 Hyundai safety: re-enable button spam safety check 4d1edc0 skip tx_hook if a message is forwarded (#379) df2ff04 bump version 168461d added fault state to health packet b3e1a13 uds: better debug prints 68c39fb uds: no need for threads if you always drain rx 91b7c5b bump Panda Ver 26cb4dc Fixed pylint error 32725cc Fixed misra compliance e33b4be Added echo script 312ba62 minor comment cleanupo e90897a Fix board detection on white 0e72c18 always stop executing if safety mode fails to be set (suggested by jyoung8607) e8d7ed1 Rename function name to not confuse safety_set_mode and set_safety_mode ff86db6 improve uds message processing 512ab3f except Exception 37ce507 py3 all bac4d85 dos and python3 501db8d uds drain before send and use has_obd() f2cbec1 Added has_obd() to python library 48e5b18 Add SDK downloading to the build step (#314) e0762c2 Add Python & USB API for controlling phone power (#313) ba9fb69 New health packet struct also in the python libs git-subtree-dir: panda git-subtree-split: a648cca
Vaggysag
pushed a commit
to Vaggysag/openpilot
that referenced
this issue
Apr 21, 2020
* Remove USB 2.0 Extension from BOS descriptor. * Declutter usb.h of unused code. Now that I'm more confident the USB 2.0 Extensions are only required for USB 3.0 devices, and there's less mystery around why the change worked for us, garbage-collect the commented-out descriptor.
nelsonjchen
pushed a commit
to LexusRXopenpilotUG/openpilot
that referenced
this issue
May 2, 2021
dkiiv
pushed a commit
to dkiiv/openpilot
that referenced
this issue
Aug 28, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Touching the gas pedale should not disengage the Adaptative Cruise Control nor openpilot with the Toyota Prius/Prime.
The text was updated successfully, but these errors were encountered: