10
$\begingroup$

Benedict Evans, a general partner at Andreessen Horowitz, claims that “almost all autonomy” projects are using lidar for SLAM, and that not all of them use HD maps.

An MIT group is testing self-driving cars on public roads without HD maps.

My question is whether the difference in error between lidar and cameras is significant. Benedict Evans and others claim that lidar is necessary for accurate enough SLAM in self-driving cars, but at first glance the KITTI benchmark data seems to contradict that claim. I want to confirm or refute that impression.

The KITTI Vision benchmark leaderboard for visual odometry/SLAM methods shows a lidar-based method called V-LOAM in first place, and a stereo camera-based method called SOFT2 in fourth place. They have the same rotation error, and a percentage point difference of 0.05 in their respective translation errors.

Is a 0.05 percentage point difference in translation accuracy large or insignificant, when it comes to an autonomous car navigation?

The KITTI Vision benchmark leaderboard for odometry/SLAM methods:

The KITTI Vision benchmark leaderboard for odometry/SLAM methods

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ Fair enough! I've rolled the comments you made here back into the question and reopened it. $\endgroup$
    – Chuck ♦
    Commented Aug 8, 2018 at 12:41

1 Answer 1

Reset to default
1
$\begingroup$

In other words, could this difference noticeably impact safety or reliability? -> not at all in my opinion.

What important in autonomous car navigation is localization stability rather than the odometry estimation accuracy. Maps for autonomous driving are usually prebuilt and globally optimized before they are used for a navigation. It is never required to build such an open loop trajectory for a navigation.

Even for a map building, 0.05% is almost meaningless if there is a proper place recognition system.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Elon Musk has recently commented that LIDAR based systems will be doomed because they are expensive while launching Tesla's SoC chip computer for Autonomous driving during Autonomy Investor Day[link]youtube.com/watch?v=bZxTG7DmB_0[link] $\endgroup$
    – Pradeep BV
    Commented May 20, 2019 at 7:51
  • $\begingroup$ He has been saying that for a long time. Well if I am buying a self-driving car I will have one with LiDAR and camera both. One of the point that he kept saying that LiDAR is not necessary was because he thinks if a camera cannot see LiDAR can't too which is not true anymore with new LiDARs. $\endgroup$
    – Chanoh Park
    Commented May 21, 2019 at 9:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.