ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

teddyyyy123's profile - activity

2017-05-15 23:35:27 -0500 received badge  Famous Question (source)
2017-05-15 23:35:27 -0500 received badge  Notable Question (source)
2016-06-28 16:17:08 -0500 received badge  Famous Question (source)
2016-05-08 14:27:08 -0500 received badge  Famous Question (source)
2016-01-08 09:34:34 -0500 received badge  Notable Question (source)
2015-12-23 09:45:33 -0500 received badge  Famous Question (source)
2015-12-22 18:46:30 -0500 received badge  Notable Question (source)
2015-12-22 16:14:00 -0500 received badge  Popular Question (source)
2015-12-22 14:00:57 -0500 asked a question Devel environment for C++?

do u guys just use emacs/vim + cscope/ctags ?

eclipse CDT would give u real-time compile/syntax check, and easier and more powerful code browsing/refactoring, would it be possible to use eclipse CDT? anybody had experience using CDT with ROS code base?

2015-12-22 08:00:39 -0500 received badge  Popular Question (source)
2015-12-22 03:12:30 -0500 received badge  Notable Question (source)
2015-12-22 03:12:14 -0500 asked a question cmd_vel or commands/velocity?

I'm trying to use the existing amcl/gmapping and create my own wheel motor drive actuator. it seems that the output from navigation (amcl/gmapping) are fed into move_base http://wiki.ros.org/move_base

the above wiki shows that move_base dumps the velocity commands to /cmd_vel

but on the turtlebot simulator, I see the commands finally go to */commands/velocity, in fact the /cmd_vel topic doesn't exist at all.

so which one is it really? I just need to know the correct place for my actuator to get its commands from.

also gmapping does SLAM so supposedly it can also do localization, what is the reason we use a separate navigation package (amcl) ?

thanks Yang

2015-12-08 22:09:11 -0500 received badge  Popular Question (source)
2015-12-08 13:44:30 -0500 commented answer occupancy grid implementation?

thanks, my goal is to pick an existing package which is the most mature and has the highest performance , so that I can just use it on my project without too much investigation and comparison by myself.

2015-12-08 13:43:31 -0500 answered a question occupancy grid implementation?

thanks, my goal is to pick and existing package which is the most mature and has the highest performance , so that I can just use it on my project without too much investigation and comparison by myself.

2015-12-07 12:19:48 -0500 asked a question occupancy grid implementation?

what is the most commonly used package in ROS for occupancy grid-based SLAM? I see slam_gmapping, octomap, voxel_grid

the jossum project also gives a very good demo

2015-12-07 00:51:54 -0500 asked a question codegen production code?

ROS separates code into separate components (nodes), and nodes communicate with each other through pub/sub.

since nodes are started from separate commands, I guess each node is a separate process, and pub/sub is implemented on top of IPC or some network service.

but IPC or network service could be slow and costly in CPU cycles, this would be a bigger problem for systems constrained on size and the computing platform they can use (for example a Raspberry-Pi type of system). ROS is great for the development process, but once debugging/tuning is done, it would be nice to compile the code or codegen into a more tightly-integrated set, maybe into a monolithic binary, and the target platform simply runs that binary, this might provide big savings in CPU and latency.

2015-11-30 22:15:42 -0500 received badge  Popular Question (source)
2015-11-30 12:11:21 -0500 asked a question what should be my recommended hardware architecture?

I'm essentially trying to build a robot platform (similar to Turtle) from scratch. part of the reason is that Turtle started many years ago, and after that, hardware technology have progressed a lot (multiple powerful ARM chips appeared , etc). my understanding is that Turtle basically uses a small desktop computer with a miniATX motherboard, sort of.

at first I tried to do all the processing on an Arduino, but then found that Arduino (at least UNO) is way too simple, doesn't have a competent OS, hence possibly can't run the full C++ code in which ROS is written in. Arduino TRE does run Linux, and is a small form-factor and consumes just 5v power supply to be very light. but I am still not sure if it's powerful enough to run complex algorithms such as video feature extraction and then SLAM.

another attempt is to run an Arduino board solely for the purpose of interacting with sensor data collection, and sending out actuator commands, and then communicate all the in/out data with another computer . the latter may be an Android phone, or a small desktop. but using a desktop requires me to provide a big power supply (how is the Turtle getting its power?)

Thanks Yang