Home | Tutorials | Wiki | Issues
Ask Your Question

John Nagle's profile - activity

2018-05-11 11:43:57 -0500 received badge  Necromancer (source)
2015-10-14 02:19:26 -0500 received badge  Taxonomist
2015-04-05 22:25:11 -0500 received badge  Famous Question (source)
2014-05-04 14:13:05 -0500 received badge  Famous Question (source)
2013-09-16 16:55:20 -0500 received badge  Famous Question (source)
2013-08-09 12:20:35 -0500 received badge  Notable Question (source)
2013-08-09 12:19:57 -0500 commented answer Gazebo being moody (sometimes starting, sometimes not)

This sort of interprocess communication trouble is why real robots (BigDog, LS3, Atlas) run the QNX operating system, where interprocess communcation Just Works.

2013-08-06 17:01:10 -0500 received badge  Notable Question (source)
2013-08-05 19:14:17 -0500 answered a question Stained STL files in Gazebo (on 64 bit system)

That sort of graphics artifact usually indicates two triangles drawn on top of each other. Look for a bug which sends the same geometry to OpenGL twice.

2013-08-05 10:20:41 -0500 received badge  Notable Question (source)
2013-08-05 07:41:09 -0500 received badge  Popular Question (source)
2013-08-02 15:19:01 -0500 asked a question Which is less awful? Gazebo/ROS or OpenHRP3/RTC builder?

I'm trying out both Gazebo/ROS and OpenHRP3/RTC for some legged locomotion theory work unrelated to the DRC.

OpenHRP3/RTC:

OpenHRP3 by itself has a solid physics simulator. I've been trying hard cases for foot-ground contact. Friction seems to work reasonably and force sensors at the ankle return values that look realistic. So you can do real legged locomotion work in there. That makes sense; that's the program used for most humanoid robots in Japan.

The GUI program for OpenHRP3 is very brittle. Do anything wrong, and some part of it crashes, or Java crashes, or Eclipse crashes, or the CORBA ORB locks up. In one case, the entire Ubuntu GUI locked up. It's very easy to cause a null pointer de-reference from the GUI. Then it's necessary to delete the history of the CORBA ORB and reboot before running the program again. This is actually documented. Sometimes the entire Eclipse workspace has to be deleted to stop repeated crashing.

RTC Builder also has a GUI, with lots of tabs full of forms to fill in. I'm stuck there; I can't get the "Data Ports" form to accept input, and have a support request in. There's a lot of persistent state in this thing; once it gets screwed up, there may be no way back without deleting the workspace.

The documentation is in Japanese (OK) and out of sync with the code (not OK).

Gazebo/ROS:

Gazebo has an inadequate physics engine, ODE, borrowed from gaming. Users have complained about unrealistic force measurements. This is well known, and Bullet, which is better, is being installed as a stopgap until the new physics engine from Mike Sherman (who gets this stuff right) is installed in 2014. There's an amusingly casual attitude towards crap physics in the Gazebo world. The documentation actually suggests guessing at inertia tensor values, resulting in values wrong by orders of magnitude. The software really should compute inertia tensors from the geometry by default. (It's not that hard. Mirtich figured this out 15 years ago, and I have code for it.) This is part of why motion in DRC simulation videos looks so unreal. One of the teams admits they got through the first round of the DRC by exploiting bugs in the physics engine.

Gazebo/ROS integration isn't really working yet. It's supposed to work in version 1.9, which was supposed to ship in "mid-summer 2013", which was last month.

The Gazebo GUI seems to work OK, but doesn't do much. The whole system is very command-line oriented and retro. There are multiple versions of the documentation, in English (OK), out of sync with each other and the code (not OK).

(I used to write simulation engines. I sold the technology to Havok a decade ago and got out of the business. So I know it doesn't have to be this bad.)

2013-07-24 08:02:43 -0500 received badge  Famous Question (source)
2013-07-24 00:02:51 -0500 commented answer DRCSIM: simulated force measurements in wrong coordinates

The displayed forces are at a hand, not a foot, correct? What's going on in foot forces as the robot sways?

2013-07-23 18:25:27 -0500 received badge  Popular Question (source)
2013-07-22 18:12:37 -0500 commented answer How to control robots models?

Re: "You should also think about using the new gazebo version(1.9)". That's not released yet.

2013-07-22 18:12:27 -0500 commented answer How to control robots models?

Re: "You should also think about using the new gazebo version(1.9)". That's not released yet.

2013-07-22 18:12:16 -0500 commented answer How to control robots models?

Re: "You should also think about using the new gazebo version(1.9)". That's not released yet.

2013-07-18 11:54:32 -0500 received badge  Teacher (source)
2013-07-16 19:52:58 -0500 commented question Videos that show Gazebo doing hard dynamics?

What I'm looking for here are results comparable to what OpenHRP3 was delivering a few years ago. OpenHRP3 has its own physics engines, developed at universities in Japan specifically for accurate robot simulation of humanoid robots. OpenHRP3 looks quite effective, but the support resources are mostly in Japanese.

2013-07-15 08:38:22 -0500 received badge  Student (source)
2013-07-15 01:02:12 -0500 answered a question Gazebo grass/pavement plane flickering

Flickering in OpenGL is generally an indication that you have two faces which are very close together, close enough that they render to the same Z-buffer value. Two instances of the same object at the same location will do it.

2013-07-15 01:00:19 -0500 received badge  Popular Question (source)
2013-07-12 11:38:00 -0500 asked a question Videos that show Gazebo doing hard dynamics?

Are there any videos that show Gazebo doing any moderately hard robot dynamics problems? Running? Hopping? Jumping? Catching? Sliding? The most "dynamic" examples I can find just show mostly-kinematic situations, such as slow walking with a bit of ZMP to keep the effective CG over the feet. The DRC videos show the Atlas robot doing a lot less than the videos of the real Atlas robot.

2013-07-10 04:00:39 -0500 received badge  Notable Question (source)
2013-07-08 02:45:32 -0500 received badge  Popular Question (source)
2013-07-04 15:38:15 -0500 asked a question How far away is ROS/Gazebo 1.9?

"Tutorials/1.9/Installing gazebo ros" Packages says: "Warning: These instructions use the latest development version of Gazebo because we are currently making a major push to revamp the ROS integration with Gazebo. Until Gazebo 1.9 is released mid summer 2013, these tutorials are considered 'Advanced' because they require you to build from source somewhat unstable code."

It's mid summer 2013. When will a released version of ROS with Gazebo be available? Is it better to wait or to struggle with "unstable code"?

2013-06-20 16:01:13 -0500 asked a question Supported graphics card: AMD Radeon HD 4250 GPU?

Is the AMD Radeon HD 4250 GPU supported for Gazebo under Ubuntu?

2013-06-19 19:28:16 -0500 received badge  Editor (source)
2013-06-19 19:26:33 -0500 answered a question What do the friction coefficients mean and why are they so large in the DRCSim Atlas URDF files?

OK. So with dContactApprox on, the friction model should be the "cone" approximated as a square-base pyramid, as shown here, right?

http://ode-wiki.org/wiki/index.php?title=Manual:Concepts#Collisionhandling

That should overestimate friction by up to sqrt(2) depending on the direction, which is probably good enough for basic legged locomotion.

If mu = mu1 = mu2 are equal, then for

|Ft| < |Fn| * mu

there should be no slippage, right?

Is all that the case, and does it work reasonably well?