GTSAM tutorial custom factor clarification.

[vsre...@gmail.com]

In the tutorial the Jacobian for the custom GPS factor is defined as

const Rot2& R = q.rotation();

 if (H) (*H) = (gtsam::Matrix(2, 3) << R.c(), -R.s(), 0.0,                                                                                                                                  R.s(),  R.c(), 0.0).finished();

However lower in the text it states that the Jacobian is

H=[100

      010]

Why is it not defined in the code as simply

const Rot2& R = q.rotation(); if (H) (*H) = (gtsam::Matrix(2, 3) << 1.0,0.0, 0.0,

                                                                                                                   0.0,1.0, 0.0).finished();

[Dellaert, Frank]

The tutorial text is wrong. The reason is because the Jacobians are defined as in math.pdf and the manifold retractions are applied in the *local* coordinate frame. I am going to shift this tutorial to a different medium soon and clean up this mistake.

--
You received this message because you are subscribed to the Google Groups "gtsam users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gtsam-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gtsam-users/064111a2-7be3-4cd3-88de-9d47c5737b27n%40googlegroups.com.

[Nicolas Perez]

There is a similar discussion in another thread: https://groups.google.com/g/gtsam-users/c/p9-rmuMa-40

[Dellaert, Frank]

I fixed the docs in PR #1139 and also copied the explanation to
gtsam.org/tutorials/intro.html#magicparlabel-65484

Please let me know if you see any bugs in the text or code. I already saw one small typo,  which I will fix before I merge.

The python custom factor example is still wrong, and I'm counting on its author (Fan!) to fix it.

Best

Frank

---

From: gtsam...@googlegroups.com <gtsam...@googlegroups.com> on behalf of Nicolas Perez <nicolasa...@gmail.com>
Sent: Friday, March 18, 2022 11:11
To: gtsam users <gtsam...@googlegroups.com>
Subject: Re: [GTSAM] GTSAM tutorial custom factor clarification.

 

To view this discussion on the web visit https://groups.google.com/d/msgid/gtsam-users/f2b0dc59-0a95-45d0-aa6d-32ab422db9e7n%40googlegroups.com.

[vsre...@gmail.com]

Hmm I think there is a problem in the equations when explaining how the Jacobian came about. Second row last element of the exponential map should be delta_y not delta_x.