CEO challenge - Unlimited XNGP ADAS full field test

[@yome1562]

Relatively new owner of a G9 in Guangzhou. With many questions/comments: - car signals a change lane with little or no time in advance. Takes you by surprise and by the time it signals it's already doing it making it difficult to check if it is appropriate or not (since you are supposed to be vigilant at all times) - Tends to drive towards the right of the lane, not in the center, even if there are cars towards the left side of the lane next to you. - Hesitates a bit when surpassing a big vehicle. Sometimes I need to intervene by accelerating to make it finish the operation. (although in my last experience it felt much better, not sure if already addressed) - I would like to learn more about how the AI works. What helps more? (as training) Using the autopilot or driving manually and then making the car learn the route (using the valet function) Guys, no complains. Very satisfied. Just would like to contribute more to make the system even better. 🎉

[@eaojnr]

I’ve raised this point so many times. Erratic signaling and restive lane changes are a continuing problem even into 5.1.2.

[@EJFJ-Corporation]

@@yome1562 Very good that you take things positively as you do... Things are getting better and better, now you are a scientist who try a fantastic tool, which is not very dangerous, but which is new and trying to do automatically what Hunan beings create seldomly and randomly... Road are not humano-normed as they should be, with few lanthanid atoms illustrating the type of road, the direction to reach the next hypo-magnetic microship allowing the vehicle to understand if the road is in a good state or damaged by even an earthquake!

[@EJFJ-Corporation]

@@eaojnr Complain as Elon Musk would do, improvements made now will allow new users to try Xpeng, Xpeng Mona, and Nio and others future Xpeng partners, allowing to quickly make them be not only better than Tesla, but the only totally secure and effective ADAS 5.0 in China and then elsewhere in the world! In Germany it seams to work very well, according to videos I saw... In China it works well in some places, but it need to be globalized by understanding which specific manner is the best in any sort of specific case! We are all scientists, trying to make things respecting alive beings most fundamental needs: security, freedom, enjoyment, doing healthy activities and realizing useful things to make life better... Animals do this, robots have to do it too and try to, because they have this legal obligation, but unfortunately human beings do not have this obligation to respect even their own most fundamental needs! At least, they are not helped to benefitting of them!

[@herrlich7671]

impressive how the system handles such complex situation @12:00

[@eaojnr]

With high definition maps from a 3rd party, why not. I get worried when the hider maps is not up to date with actual road conditions. In those situations what happens is the car then drops to LCC

[@herrlich7671]

really good to see that also the CEO exactly knows whats going on!

[@mickjoebills]

Which level of xpilot is available in new G6? in Australia?

[@EJFJ-Corporation]

Very good presentation, and not like Tesla which works very correctly only on some specific roads or in some conditions. Xpeng is applying global solutions made for the global world, with lot of different situations which should be respected and used correctly, as local specificities but which can be transposed positively in other similar conditions or when this method is convenient in a strange situation, to respect road users habits and surprise them conveniently (without taking any risk)... It needs to respect all users, and not only the bad driver's habits and surprise him positively as a professional taxi driver will do! But it allows to respect the wishes of all road drivers, soaring risks, and making all not taking any bad habits d'y doing things as an Amator driver and not as a conscious and efficient driver, who spares both tyres damages and energy wastings! Things needs to be considered both globally but accurately in this very specific and unique situation!

[@kimdr.5231]

Xngp is not for some specific road, its all of chinese road

[@EJFJ-Corporation]

@@kimdr.5231 Yes, exactly what I said, but it is now adapted in different countries, to respect their local wasteful and illicit rules to drive not very efficiently... They put lot of different expensive but inefficient speed limitations systems, that increase néfaste disparities between those who have strong and efficient vehicles abd those who have old ones or sportive ones ... Xpeng Motors does very efficient vehicles which can adapt fully to the conditions, but they need to learn how things work locally... They now use good ways to improve, by storing what human beings propose, as it should be globally done by EJFJ Positive Contextualisation since March 15th, 1996!

[@EJFJ-Corporation]

@@kimdr.5231 Yes, thank you, my sentence was not clear, I corrected it... Tesla is not made globally, it is tried in California and sold in the global world, but with only legal adaptations, but not the most secured ones like not letting anything touch the car!

[@eaojnr]

@@kimdr.5231 not true! Stop purporting misleading information. xNGP is not for all roads. Check their release notes.

[@eaojnr]

Nope, you're misleading folks. Tesla is using pure vision, so it learns on-the-go and adapts. It doesn't rely on high-definition maps, and that's the reason FSR works correctly in any road circumstance with little helper data, like high-definition maps, or additional telemetry data.

[@petitbarzun]

Thank you! When do you expect NGP for highways in Europe? I am eagerly awaiting my G6 in the Netherlands.

[@EJFJ-Corporation]

@@petitbarzun In Germany, they just arrived but during tests it seams it works already much better than opponent systems. In the Netherlands, it should be the case, especially if you receive your vehicle soonly...

[@petitbarzun]

@@EJFJ-Corporation It has very good Adaptive Cruise Control, Lane Centering Control and Adaptive Lane Change. In China, the G6 can also do this based on navigation. You tell it where to go and it will choose an appropriate lane for its speed and take an exit on the highway, without user interaction. As shown in the video, it can also do that in cities in China, but I guess that will take longer to bring to Europe.

[@mickaelchartier2254]

My dream car 😍😍 looking for one but to expensive for me 😭😭

[@gkdotclass]

We need Xpeng in Portugal!

[@rj8u]

When the orange car was stopped on your lane the system should have stop and waited for the opposite car to pass first because they have the right if way then it should have proceeded after instead of cutting the double yellow lane in my opinion.

[@TariqZahid-d1l]

Why he is keeping hand on the steering, it must be demonstrated without touch the steering.

[@EJFJ-Corporation]

@@TariqZahid-d1l No, the steering wheel is useful to put hands on, it is not nefast in anything, he does not drive, he follows... But if the driver or the front passenger fully blocks the driving wheel, the car should drove normally, conveniently, sparing any accident. And even the driver turns the driving wheel totally incorrectly, the vehicle needs to follow an existing road and protecting all users. The vocal system should ask the car inoccupants what they want to do, take a rest on a secure border side, move quickly to a hospital, and help enjoyment of all users...

[@eaojnr]

The car can make erratic moves sometimes. So keeping hands on the steering is a required safety initiative.

[@eaojnr]

I’ll be releasing more videos soon as a real owners experience for routes that have not been highly optimized for these marketing videos

[@EJFJ-Corporation]

@@TariqZahid-d1l Others alternative are arriving, and from Xpeng, not from other brands... We will see if Nio adopts XNGP too and collaborates improving it too, as they did a quote correct one, but much less complete and less efficient than the very fantastic XNGP ..

[@EJFJ-Corporation]

@@eaojnr This needs to be solved! Thank you for requesting this whenever you can to Xpeng staff or technical support ... I hope they use Tianjin OS on their system and that they do not have to repeat to engineers what they read, and that AI concretely help them improving things and show their habilités rather than having to write a lot and search for little errors! 😉

[@LouiseViolet-q9n]

Rosa Haven

[@ReidRobinson-r6q]

McLaughlin Springs

[@LoiseKurschner-e2h]

Daniel Lodge

[@AnetteBoshears-p5d]

Greenholt Forge

[@RodolfoLatham-e5y]

Araceli Ridges

[@WallaceBennett-k8l]

Miller Overpass

[@TerryCrosby-m9s]

Osborne Well

[@ZimmermanMark-l6d]

Goodwin Inlet

[@KevinBurnett-h6v]

Glenna Field

[@RodneyDail-o2u]

Jast Forks

[@JoelHyman-b2c]

Lubowitz Valleys

[@HansomAdelaide-i4q]

Richie Valleys

[@MacaulayDuke-l7k]

Lowe Field

[@NickMonroe-u7f]

Axel Squares

[@EdenJane-p6c]

Oberbrunner Vista

[@AngelaHenderson-q5w]

Schowalter Point

[@PatrickMarlon-q4y]

Sean Trail

[@DeangeloMedlen-t2x]

Yundt Walks

[@FlorenceMetcalf-l3f]

Durgan Island

[@MarionEdison-b8u]

Barrows Tunnel

[@DeltaDawson-k5h]

Ike Turnpike

[@CassandraKinney-r6l]

Cicero Burgs

[@BrantHemminger-w5p]

Abbigail Well

[@MaishaLazzara-h8z]

Vaughn Bridge

[@SimonDylan-t5e]

Miller Square

[@AlmetaWilliam-z3j]

Grant Plain

[@MalanClement-e5m]

Kirlin Brooks

[@VannessaCollison-w4s]

Ortiz Via

[@TuiuioGhujko-m7c]

Greenfelder Manor

[@WilbertSimon-j7b]

Dare Overpass

[@JamesVanscoy-k6n]

Hegmann Lodge

[@JeanJohn-f9i]

Elfrieda Valley

[@AndreaAdams-j3h]

Laury Island

[@ZephaniahBradley-s1y]

Boyle Dale

[@HerbertBeau-z4r]

Martinez Frank Garcia Jessica Davis Karen

[@JuliusYehudi-l7c]

Issac Parks

[@RogerCalvin-q5p]

Roob Road

[@BlackMaureen-y9z]

Lurline Fork

[@StevenMartin-g3u]

Ziemann Island

[@ToshikoGasiewski-s1o]

Gerhold Heights

[@KentLucy-n4j]

Mohr Point

[@ValeryHarmon-g4o]

Treutel Center

[@AugustusMatt-t7w]

Prohaska Row

[@DarronHaig-q7b]

Berta Ferry

[@MoraEroh-n5n]

Nicholaus Drive

[@ArnoldMarlon-t5j]

Destinee Mills

[@KyleSells-c2p]

Gus Tunnel

[@ReadeNovia-x2e]

Austen Freeway

[@JanettaGarcia-x1w]

Rhianna Neck

[@LidaYilma-y1m]

Nicola Square

[@gadgetgasspoll]

X-Punk

[@SokinaAktar-g8v]

Harris Edward Davis Michelle Martin Angela

[@JuniorRingeisen-x8v]

Sydnie Plain

[@CynthiaMiller-n8v]

Mayer Hollow

[@WycliffeSherry-o2m]

Kelsie Villages

[@LawrenceVarquez-q1c]

Waelchi Hollow

[@KirkQuam-q9k]

Fae Loop

[@RoseClare-s6o]

Harmony Union

[@PatriciaHaley-z6d]

Ena Isle

[@AgnesMirabelle-j2n]

Perez Gary Wilson Mary Hall Dorothy

[@EffieLeatherman-s9s]

Champlin Station

[@FrancesMcken-e2k]

Destiney Meadows

[@kumbackquatsta]

and elon is delaying the robotaxi once again

[@JamesSeyler-u6z]

Robinson Jeffrey Garcia Kimberly Davis Nancy

[@HoodDuncan-b6u]

Gayle Key

[@BethuneLowell-l3k]

Halie Unions

[@ElsieMonroe-r2n]

Zemlak Port

[@fzhxns-h5w]

Fay Road

[@DanielRuff-n9x]

Konopelski Fall

[@RockefellerYetta-h8l]

Walter Islands

[@DanGlover-b8c]

Nikolaus Unions

[@DomenicAldrich-q8w]

Gerlach Union

[@VoppedsTonns-n6h]

Myrna Flat

[@eaojnr]

Road complexity in US is more challenging than China. Don’t make that mistake of thinking it upside down.

[@petitbarzun]

I have never seen the road complexity in the US as complex. Europe is easy more complex and Asia is even worse

[@eaojnr]

@@petitbarzun and I agree to that too. Comparatively Europe is very complex. Imagine London roads… it’s a nightmare. But in order of difficulty, I’d say US is more complex than China given that road strictness is not as high as China. In China road strictness is so high that it is relatively easier having the autopilot stick to policies implemented in its rules engine. For instance no overtaking on continuous line, or no overtaking at school areas, traffic lights count down aided by 3rd party integration and so many more road aides in China. You don’t get that in US.

[@xtreamerpt7775]

@@petitbarzun as a european that went on vacation to the US i have to agree.

[@EJFJ-Corporation]

@@eaojnr Do not mistake mess and complexity! In USA there are everything, even poor persons who want to stop suffering by making others suffering as much as they can... In Asia, they are far more respectuous of human lives, and do not do anything nefast. But as there are far more road users each hour in one place, and as many are thinking of problems they life outside of their car when they are staying in their car, they can make mistakes or take inconsiderated risks! So, this is complexity, to manage very different users having different needs ... In Europa, there are lot of different situations, each round about is specific, each country and each city have specific way to treat things, specific rules that only the local drivers know ... There are far too many informations, not all can be seen by driving... Europa exploits risks management and néfaste disparities to make money turn from one citizen to others...

[@eaojnr]

@@EJFJ-Corporation Mess is part of the complexity to manage through perceptive receptors of autonomous robots. Complexity is a myraid of concerns, including marked and unmarked road interests. If roads were clean, with very good lane markings, well spaced, with strict road use enforcement, you can have a more managed capability to handle perception and planning. In the absence of that, you have to manage a more complex scenario. This is the reason it is completely misleading to say China roads are more complex than US. I've lived in both places, and it's almost a naive statement. In China, Xpeng as an example sources high-definition maps, and roads are very well demarcated most of the time. In the absence of well-demarcated roads, there are good kerbs and good trafic monitoring systems in place to check bad road use, and on top of all that, training schools for drivers ensure that drivers adhere strictly to road regulations and stipulations. You can't get that in the US. You have road rage, uncontrolled walk across streets, some narrow roads and so much more mess to have to handle. Now Tesla doesn't use LiDAR, but only relies on pure camera, and yet, it does much better based on its perception pipeline. If you take a Tesla FSD and put it on a Chinese Road it will excel almost instantly given the more orderly state of roads in China. If you put Xpeng latest XOS in US it will fail because it doesn't have high-definition maps. The only way you will be able to use it favorable ok, will be with the "AI-Driver" feature, which is a clever workaround to not having high-definition maps. AI-driver is another integration that leverages waypoints (Refer to DJI Drone Waypoints) to do precision recording of the path, and then you can save it and play it over and over again. Am i chatting with a XPENGs bot - go figure. lol!

[@JaneToby-r3j]

Haley Light

[@GeorgeSmith-v1p]

Casimer Dam

[@LawsonClarence-i3f]

Oleta Islands