Quaternion convention about ethzasl_msf HOT 5 CLOSED

Ok, so let's start the discussion ;)
As most of you know, I'm a high supporter of Hamilton,

• Eigen uses it
• ROS uses it
• In my opinion, transformations are more intuitive
• Just fixed / checked / changed everything to Hamilton for my thesis
• With the definition of the error in msf, the error terms in the Jacobians end up being the same anyway

from ethzasl_msf.

My "unbiased" vote goes for JPL notation ;-)

Jokes aside. I am surprised that sm and aslam use JPL notation, given the efforts "we" did to change ssf and msf to Hamilton. I think there is no need to figure out why they used JPL notation but I feel we should go with the Hamilton notation - just because everyone else does. To my knowledge JPL papers are so far the only ones that do not use Hamilton.

From: Markus Achtelik [[email protected]]
Sent: Friday, August 09, 2013 1:42 AM
To: ethz-asl/sensor_fusion
Subject: Re: [sensor_fusion] Quaternion convention (#19)

Ok, so let's start the discussion ;)
As most of you know, I'm a high supporter of Hamilton,

• Eigen uses it
• ROS uses it
• In my opinion, transformations are more intuitive
• Just fixed / checked / changed everything to Hamilton for my thesis
• With the definition of the error in msf, the error terms in the Jacobians end up being the same anyway

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/19#issuecomment-22382396.

from ethzasl_msf.

For what it's worth, google ceres uses the hamiltonian convention too..

On 09.08.2013, at 18:45, stephanweiss [email protected] wrote:

My "unbiased" vote goes for JPL notation ;-)

Jokes aside. I am surprised that sm and aslam use JPL notation, given the
efforts "we" did to change ssf and msf to Hamilton. I think there is no
need to figure out why they used JPL notation but I feel we should go with
the Hamilton notation - just because everyone else does. To my knowledge
JPL papers are so far the only ones that do not use Hamilton.

From: Markus Achtelik [[email protected]]
Sent: Friday, August 09, 2013 1:42 AM
To: ethz-asl/sensor_fusion
Subject: Re: [sensor_fusion] Quaternion convention (#19)

Ok, so let's start the discussion ;)
As most of you know, I'm a high supporter of Hamilton,

• Eigen uses it
• ROS uses it
• In my opinion, transformations are more intuitive
• Just fixed / checked / changed everything to Hamilton for my thesis
• With the definition of the error in msf, the error terms in the Jacobians
  end up being the same anyway

—
Reply to this email directly or view it on GitHub<
https://github.com/ethz-asl/sensor_fusion/issues/19#issuecomment-22382396>.

—
Reply to this email directly or view it on
GitHubhttps://github.com//issues/19#issuecomment-22407678
.

from ethzasl_msf.

This depends on which Quaternion parameterization you pass for the respective parameter blocks :)

From: Sammy Omari [[email protected]]
Sent: Friday, August 09, 2013 18:56
To: ethz-asl/sensor_fusion
Cc: Lynen Simon
Subject: Re: [sensor_fusion] Quaternion convention (#19)

For what it's worth, google ceres uses the hamiltonian convention too..

On 09.08.2013, at 18:45, stephanweiss [email protected] wrote:

My "unbiased" vote goes for JPL notation ;-)

Jokes aside. I am surprised that sm and aslam use JPL notation, given the
efforts "we" did to change ssf and msf to Hamilton. I think there is no
need to figure out why they used JPL notation but I feel we should go with
the Hamilton notation - just because everyone else does. To my knowledge
JPL papers are so far the only ones that do not use Hamilton.

From: Markus Achtelik [[email protected]]
Sent: Friday, August 09, 2013 1:42 AM
To: ethz-asl/sensor_fusion
Subject: Re: [sensor_fusion] Quaternion convention (#19)

Ok, so let's start the discussion ;)
As most of you know, I'm a high supporter of Hamilton,

• Eigen uses it
• ROS uses it
• In my opinion, transformations are more intuitive
• Just fixed / checked / changed everything to Hamilton for my thesis
• With the definition of the error in msf, the error terms in the Jacobians
  end up being the same anyway

—
Reply to this email directly or view it on GitHub<
https://github.com/ethz-asl/sensor_fusion/issues/19#issuecomment-22382396>.

—
Reply to this email directly or view it on
GitHubhttps://github.com//issues/19#issuecomment-22407678
.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/19#issuecomment-22408323.

from ethzasl_msf.

U. Minnesota uses JPL

From: stephanweiss [[email protected]]
Sent: Friday, August 09, 2013 18:45
To: ethz-asl/sensor_fusion
Cc: Lynen Simon
Subject: Re: [sensor_fusion] Quaternion convention (#19)

My "unbiased" vote goes for JPL notation ;-)

Jokes aside. I am surprised that sm and aslam use JPL notation, given the efforts "we" did to change ssf and msf to Hamilton. I think there is no need to figure out why they used JPL notation but I feel we should go with the Hamilton notation - just because everyone else does. To my knowledge JPL papers are so far the only ones that do not use Hamilton.

From: Markus Achtelik [[email protected]]
Sent: Friday, August 09, 2013 1:42 AM
To: ethz-asl/sensor_fusion
Subject: Re: [sensor_fusion] Quaternion convention (#19)

Ok, so let's start the discussion ;)
As most of you know, I'm a high supporter of Hamilton,

• Eigen uses it
• ROS uses it
• In my opinion, transformations are more intuitive
• Just fixed / checked / changed everything to Hamilton for my thesis
• With the definition of the error in msf, the error terms in the Jacobians end up being the same anyway

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/19#issuecomment-22382396.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/19#issuecomment-22407678.

from ethzasl_msf.