Zoom Logo

Deep Dive on SLAM: Simultaneous Localization And Mapping - Shared screen with speaker view
Riael
13:09
#42gether
Max Senges
20:27
Hi everyone — looking forward to your questions and links here
Max Senges
21:06
Cyrills website https://www.ipb.uni-bonn.de/
Max Senges
26:00
If you are interested in autonomous driving, software automotive and mobility please have a look at the 42 Specialization SEA:ME we are developing and offering in Wolfsburg https://medium.com/42-wolfsburg/from-mechatronics-to-software-42-wolfsburg-gathering-open-expert-community-to-shape-future-of-6a611acf3056
42Kocaeli
28:14
:) ok
Hanifi "hanifastic" ŞİŞMAN
28:28
thx
Florence | flormich
33:31
Cub3D!
enesozmert42KOCAELI
34:59
Isn't the idea of ​​'SLAM' close to impossible in the ever-changing world? Change can be thought of as structure, continent, and idea.
cdahlhof
36:56
if the mapping "robots" were not Held in Position by Ground friction; could external Sources of disturbance "warp" the resulting map
cdahlhof
37:05
?
Mert Barut | mbarut
46:56
Is this Bayesian estimation?
atabiti 1337
50:51
What is difference between localization and mapping, and what are the common Challenges with SLAM?
nnakarac (pon)
54:39
For the RF based localization, is that considered as the part of SLAM?
cdahlhof
57:05
why are the nodes based on the time instead of space?
nnakarac (pon)
58:01
like location from GPS, mobile basestation, or other RF devices (WiFi, XBee, etc)
Dustin
59:19
RF devices = radio-frequency devices
Panyawat Rattana
01:06:56
In loop closure, How does the robot know that they have passed that position because each point cloud from sensor is not on the same exact position?
adaifi
01:11:54
how can we apply the SLAM in the AR experience
Valentin Fuhlenbrok | vfuhlenb
01:12:25
Could you repeat the name of the Github repository please?
Cyrill Stachniss
01:12:41
SuMa
Riael
01:12:49
What is the benefit of using LiDAR instead of 3D cameras like other cars are opting for?
Cyrill Stachniss
01:12:50
PRBonn
cdahlhof
01:13:45
https://github.com/jbehley/SuMa
Max Senges
01:19:49
;-)
Max Senges
01:20:23
Recommend Paris Texas: https://en.wikipedia.org/wiki/Paris,_Texas_(film)
cdahlhof
01:24:00
if each node has an error range for the next nodes-position, how do you reslove "circles" of Offset nodes?
cdahlhof
01:29:51
k
Louis
01:35:56
if you generate a loop closure constraint. but then you collect further nodes that suggests that it actually wasn't a closure (eg you realise that there are two Eiffel towers and this would be the more consistent solution) does the mapping then "unfold" and remove the loop closure or is it only forward looking and the past nodes are now fixed?
archibald 1418
01:36:36
Wow this is really illustrative and visually comparable to least squares estimation
Louis
01:38:18
brilliant thanks!
Max Senges
01:41:04
Dear Cyrill this is plain awesome!! Unfortunately I have a meeting at 3:30 so will have to leave. Lets sync 1:1 again soon
Louis
01:44:57
so is the idea for semantic slam is that you add "what it is" flags to your data and then use that additional information to judge the data's "importance". Eg if you tree leaves move, you won't give it much significance and assume it's noise. But if you see a wall move then that is important and should be included
cdahlhof
01:45:41
is it possible to have multiple "robots" to update the same map?
archibald 1418
01:47:28
How detailed the geometric semantics should be? Do we have to include eg constructive solid geometry or a simple contrast map should suffice?
cdahlhof
01:49:16
if we use this mapping System to recognize and differenciate objects in our Environment; and use this data to train an AI for the same Task, do you think the AI would surpass / outperform the SLAM System eventually?
Louis
01:50:57
how do you do your semantic segmentation? Is this by postprocessing the same lidar 3d point cloud to identify objects. Or is this some sort of sensor fusion integrating other data (eg a colour camera and computer vision or something)
Лена Логунова
01:50:57
what equipment do you use in practice for experiments and test algorithms for the movement of mobile robots or vehicles?
Yana D. | ydimitro
01:53:34
Adaifi asked before about SLAM in AR, can you tell us more about applying SLAM in AR/XR?
Louis
01:53:43
are you referring to GPS RTK or is that still not good enough as a ground truth?
cdahlhof
01:57:10
in your previous mapping visualizations the Point of origin was a blue car; was that car part of the map itself?
Louis
01:57:38
Sorry to repeat it but I think you missed it further up:how do you do your semantic segmentation? Is this by postprocessing the same lidar 3d point cloud to identify objects. Or is this some sort of sensor fusion integrating other data (eg a colour camera and computer vision or something)
gvitor-s
01:59:27
Most of the videos has only static elements ( parking cars, trees). What is the robot behavior to avoid colide against dinamics elements (eg fast cars )?
Louis
01:59:55
wow, sounds like a lot of effort went into that. Thanks!
Dustin
02:00:41
is it possible to equip cities with thousands of traffic sensors which can support the SLAM systems within autonomous cars ?
gvitor-s
02:00:54
thanks
Tetiana Fedorenko | tfedoren
02:01:07
How do you think, theoretically, who will be responsible for the possible auto crash in the case of fully autonomous driving in the future? Manufacturer, insurer, or still the driver?
Tetiana Fedorenko | tfedoren
02:03:54
Thank you very much! )
gvitor-s
02:04:07
Thank you very much !
Valentin Fuhlenbrok | vfuhlenb
02:04:18
Thank you Cyrill, it was a great talk!
virun j
02:04:25
Thank you so much
Hanifi "hanifastic" ŞİŞMAN
02:04:26
See u
Riael
02:04:26
Thank you very much for this! The SEA:ME looks amazing!
cdahlhof
02:04:27
thankys
Dustin
02:04:28
thx. mindblowing technology!
Yana D. | ydimitro
02:04:30
thank you! I really enjoyed your presentation!
Madis Luik
02:04:34
Thank you! Bye!
Tetiana Fedorenko | tfedoren
02:04:43
Thank you! Bye)
Ricardo Yoshio Yamachita | ryoshio-
02:04:54
thank you bye