Chinese self-driving car companies agree with Elon now? They are dropping use of HD maps for autonomous vehicles

firsttruck

Well-known member
Joined
Sep 25, 2020
Threads
172
Messages
2,542
Reaction score
4,042
Location
mx
Vehicles
none
Country flag
Over 7 years ago, Elon Tesla said that doing autonomous vehicles using high-definitions maps was impractical and too expensive. Looks like some companies are finally giving up using high-definitions maps.

----------------------------------

Huawei scraps expensive precision maps in self-driving cars
Huawei and other Chinese self-driving car companies are developing alternative technologies to high-definitions maps as the precision required for autonomous vehicles makes the mapping process prohibitively expensive.
By TAKASHI KAWAKAMI, Nikkei staff writer
April 26, 2023
https://asia.nikkei.com/Spotlight/E...expensive-precision-maps-in-self-driving-cars

.....
"High-definition maps are very expensive," said Huawei senior executive Yu Chengdong at Auto Shanghai show on April 16. "We have been collecting data in Shanghai for one or two years, but we still have not been able to cover all 9,000 kilometers of the city's roads."

HD maps for autonomous driving require technology more advanced than those for standard navigation systems. Maps accurate to about 10 meters are sufficient for navigation systems, but accuracy to about 20 centimeters, and in some cases as low as 1 cm, is necessary for self-driving cars, according to a Beijing research company.

The survey cost for a map with 10-cm accuracy is 10 yuan ($1.44) per kilometer, but for accuracy to 1 cm costs up to 1,000 yuan ($144) per kilometer, according to a 2021 white paper by industry group China Industry Innovation Alliance for Intelligent and Connected Vehicles.

----------------------------------
Sponsored

 
Last edited:

BayouCityBob

Well-known member
Joined
Apr 14, 2022
Threads
21
Messages
440
Reaction score
1,419
Location
Texas
Vehicles
2018 Model 3
Country flag
I think there will be a compromise. Think about human drivers for a second. We rely heavily on our maps to drive. How well do you drive in a city you have never seen before, in the middle of rush hour, for the first time? You drive a lot like Tesla FSD, because for Tesla FSD every time it drives it is discovering the landscape for the first time.

I expect there will be a compromise between the LiDAR HD maps and the very high level Google-type maps. Something that knows more about the route you are traversing so it is "familiar" as it would be to human driver but it is not HD which is unnecessary with a camera-based system.
 

Crissa

Well-known member
First Name
Crissa
Joined
Jul 8, 2020
Threads
126
Messages
16,227
Reaction score
27,092
Location
Santa Cruz
Vehicles
2014 Zero S, 2013 Mazda 3
Country flag
I think there will be a compromise. Think about human drivers for a second. We rely heavily on our maps to drive. How well do you drive in a city you have never seen before, in the middle of rush hour, for the first time? You drive a lot like Tesla FSD, because for Tesla FSD every time it drives it is discovering the landscape for the first time.

I expect there will be a compromise between the LiDAR HD maps and the very high level Google-type maps. Something that knows more about the route you are traversing so it is "familiar" as it would be to human driver but it is not HD which is unnecessary with a camera-based system.
HD maps are way beyond human memory, tho. They were trying to get around weak object recognition in the vehicle by making maps for them to have that knew where every leaf and sign and pothole was.

-Crissa
 

BayouCityBob

Well-known member
Joined
Apr 14, 2022
Threads
21
Messages
440
Reaction score
1,419
Location
Texas
Vehicles
2018 Model 3
Country flag
HD maps are way beyond human memory, tho. They were trying to get around weak object recognition in the vehicle by making maps for them to have that knew where every leaf and sign and pothole was.

-Crissa
Agree. That is why I think "higher" definition maps with more land mark recognition is needed for a camera system but I do not think that the HD maps is likely to deliver a viable path forward for anything except geofenced robot axis (where they still may make a lot of sense).
 

Crissa

Well-known member
First Name
Crissa
Joined
Jul 8, 2020
Threads
126
Messages
16,227
Reaction score
27,092
Location
Santa Cruz
Vehicles
2014 Zero S, 2013 Mazda 3
Country flag
Agree. That is why I think "higher" definition maps with more land mark recognition is needed for a camera system but I do not think that the HD maps is likely to deliver a viable path forward for anything except geofenced robot axis (where they still may make a lot of sense).
Yeah. There needs to be some mapping, and definitely you can drive better knowing what is around the corner, or when to slow down for potholes,

But actually mapping the centimeters is way too much.

-Crissa
 


CyberGus

Well-known member
First Name
Gus
Joined
May 22, 2021
Threads
67
Messages
5,817
Reaction score
19,093
Location
Austin, TX
Website
www.timeanddate.com
Vehicles
1981 DeLorean, 2024 Cybertruck
Occupation
IT Specialist
Country flag
Yeah. There needs to be some mapping, and definitely you can drive better knowing what is around the corner, or when to slow down for potholes,

But actually mapping the centimeters is way too much.

-Crissa
Right, mapping ALL of the centimeters is overkill. However, if my Cybertruck just blasted over a pothole, then it can send the high-res imagery to the Mothership for other drivers.
 

JBee

Well-known member
First Name
JB
Joined
Nov 22, 2019
Threads
18
Messages
4,752
Reaction score
6,129
Location
Australia
Vehicles
Cybertruck
Occupation
. Professional Hobbyist
Country flag
Tesla needs to do active dynamic mapping, that updates the base maps already on file.

But...I believe CT will have adaptive suspension vision, meaning like some Mercs, it will reconfigure your suspension to deal with an incoming bump that it sees ahead with it's cameras.
 

Gurule92

Well-known member
Joined
Sep 2, 2021
Threads
147
Messages
3,271
Reaction score
6,646
Location
Colorado Springs
Vehicles
MYP
Occupation
"Cyber" stuff
Country flag
we need more cars on the road so we can get out hive mind on.

Who needs maps when you can just ask your buddy.

 

HaulingAss

Well-known member
First Name
Mike
Joined
Oct 3, 2020
Threads
9
Messages
4,495
Reaction score
9,476
Location
Washington State
Vehicles
2010 F-150, 2018 Model 3 Perform, FS Cybertruck
Country flag
Over 7 years ago, Elon Tesla said that doing autonomous vehicles using high-definitions maps was impractical and too expensive. Looks like some companies are finally giving up using high-definitions maps.

----------------------------------

Huawei scraps expensive precision maps in self-driving cars
Huawei and other Chinese self-driving car companies are developing alternative technologies to high-definitions maps as the precision required for autonomous vehicles makes the mapping process prohibitively expensive.
By TAKASHI KAWAKAMI, Nikkei staff writer
April 26, 2023
https://asia.nikkei.com/Spotlight/E...expensive-precision-maps-in-self-driving-cars

.....
"High-definition maps are very expensive," said Huawei senior executive Yu Chengdong at Auto Shanghai show on April 16. "We have been collecting data in Shanghai for one or two years, but we still have not been able to cover all 9,000 kilometers of the city's roads."

HD maps for autonomous driving require technology more advanced than those for standard navigation systems. Maps accurate to about 10 meters are sufficient for navigation systems, but accuracy to about 20 centimeters, and in some cases as low as 1 cm, is necessary for self-driving cars, according to a Beijing research company.

The survey cost for a map with 10-cm accuracy is 10 yuan ($1.44) per kilometer, but for accuracy to 1 cm costs up to 1,000 yuan ($144) per kilometer, according to a 2021 white paper by industry group China Industry Innovation Alliance for Intelligent and Connected Vehicles.

----------------------------------
It's interesting how many years ahead of everyone else Elon already was, simply by using logical thinking! This is why Tesla is able to focus on solutions like a precision laser and almost all the rest struggle to make a single dollar on EV's. We were told by the automotive "experts" that the OEM's would "eat Tesla for lunch" when they decided to make EV's (due to their 100 plus years of automotive experience) but that perspective couldn't have been more wrong.

It turns out that thinking logically is actually more important than doing things the same way they have always been done!
Sponsored

 
 




Top