Driver and USBTraq Alternatives for Linux systems (or ROS packages)

Moderator: mikeh9

kevin0932
Posts: 6
Joined: Fri Feb 16, 2018 10:51 am

Driver and USBTraq Alternatives for Linux systems (or ROS packages)

Postby kevin0932 » Tue Apr 03, 2018 9:25 pm

Dear Indotraq team and users,

I wonder if there are any alternative software and device drivers tailored for the Linux system (Ubuntu 14.04 or 16.04)? Currently I only find the driver for Windows systems, and Termite/USBTraq seem to only work with Windows.

For our robot localization application, real-time localization data is required and we prefer the tools compatible with Ubuntu systems. It would be the best if there are any existing interfaces for Robot Operating System (ROS) development. Do you have any experience on getting real-time localization feedback of the tag being tracked? For us, an online tracking data communication should be established.

Thanks for sharing all the information.

Best regards,

Kevin
mikeh9
Site Admin
Posts: 29
Joined: Thu Mar 17, 2016 4:39 pm

Re: Driver and USBTraq Alternatives for Linux systems (or ROS packages)

Postby mikeh9 » Wed Apr 04, 2018 1:02 pm

We use STM32F4 chips. STMicroelectronics only provides PC based virtual comm port drivers, but by searching the web, I find quite a few people using these chips with Ubuntu. As an example, see the following link:

https://electronics.stackexchange.com/questions/283852/stm32-usb-cdc-firmware-to-autodetect-by-linux
kevin0932
Posts: 6
Joined: Fri Feb 16, 2018 10:51 am

Re: Driver and USBTraq Alternatives for Linux systems (or ROS packages)

Postby kevin0932 » Sun Apr 08, 2018 4:41 pm

Thanks a lot, @mikeh9.

As Indicated by your link, I tested in Ubuntu 16.04 system where usb_cdc_acm module is normally loaded. In case that anyone else is interested, here I confirm that after connecting the device via USB and turning it on, it is automatically detected and recognized by Ubuntu system as, say, /dev/ttyACM0. You can easily get the data output by reading from the corresponding serial port. Also, the configuration commands can be sent to the device by writting to the port. As for the ROS interface, I will check out the rosserial library later or some interface wrapper can be done.

I have more questions about the values I got with automatic calibration and I am still confused about the min. offset value in the configuration.

In USBTraq, there is a field "Min. Offset value" under the tab "Anchor Positions" and it is supposed to be the offset value added to the z-axis of the anchor positions. Would you please confirm which data outputs will be affected by this offset? e.g. the z values in the output commands "$, 10006" and "$, 10001"? Will it affect the raw range data output in the commands "$, 10002" and "$, 10003"?

In other words, are the range data (in mm) from the commands "$, 10002" and "$, 10003" are supposed to be very close to those I measured manually between the anchors or between the tag and the anchors? As shown in the attached figures (from a rectangular setup with 8 anchors), the z-values of anchor positions and tag positions are offset with 600, while the range data from tag to anchors are different from what I expect. The x and y values of anchor positions in the command output "$, 10006" indicates that the width (1993 in y-axis), the length (2662 in x-axis) and the height (2176-600=1576 in z-axis) of the rectangular configuration (while the real setup is width=1.8m, length=2m and height=1m). The figure shows the results I got when I placed the tag very close to the master anchor (id=0) and the range data from the tag to the master anchor still gives me 640mm. For me, it seems there are still some points I missed or setup it in a wrong way. Generally, I didn't do any manual calibration and auto-calibration with anchor 4 is below anchors 0-3 is enabled. Do you have any suggestions on how to get better accuracy?

Image
(please find the figure here https://github.com/kevin0932/misc/blob/ ... estion.png)


Also, indicated by the specifications, Distance Precision is about ±5 mm. When converting it to tag position precision, shall I expect the localization accuracy to be within 1 cm when the system is correctly setup, say 3m by 3m as the same configuration is used in the report (http://indotraq.com/wp-content/uploads/ ... curacy.pdf)? Generally, will the system be affected if there are occlusions like human in the tracked 3D space or there are 5GHz wifi deployed in the same space?

I am quite new to the system and the technology behind. Really appreciated for your time and great help.

Best,

Kevin
mikeh9
Site Admin
Posts: 29
Joined: Thu Mar 17, 2016 4:39 pm

Re: Driver and USBTraq Alternatives for Linux systems (or ROS packages)

Postby mikeh9 » Mon Apr 09, 2018 12:50 pm

The minimum offset sets the vertical positions of the bottom anchors for the automatic anchor positioning mode. Therefore it will also affect the 10006 command which are the anchor positions and 10001 command which contains the tag position.

Commands 10002 and 10003 will not be affected by the minimum offset. Calibration if needed for your application will improve both 10002 and 10003 measurements.

The actual location accuracy you get will depend mostly on calibration, but will also be affected by whether the tag is inside or outside the tracking space. A human blocking one tag to anchor line of sight with 8 anchors will have some affect, but it will be less with 8 anchor system because the other tag to anchor measurements will make up for the human occlusion.

Other wireless devices will not affect the Indotraq system because it operates at completely different RF frequencies than things like cell phones, 2.4G wifi, Bluetooth or 5G wifi. The only problem you might encounter is if you use the system in very small rooms. The reason is the wireless signal will reflect off all the walls and combine to create a noisy signal. The Indotraq system is really best suited for large spaces.

Please send email to support@indotraq.com for more detailed support with your application.
kevin0932
Posts: 6
Joined: Fri Feb 16, 2018 10:51 am

Re: Driver and USBTraq Alternatives for Linux systems (or ROS packages)

Postby kevin0932 » Mon Apr 09, 2018 2:03 pm

@mikeh9, thank you so much for your explanations. The answers are really helpful!

The final deployment will be done in a room of size about 6m*6m*5m so it won't be really small. Then I will try to reduce human occlusion and do some calibration to see if I can increase the accuracy. Later we will also try to use the system outdoor.

I am really appreciated for your help and will write emails if we encounter further problems. Thanks!

Return to “Software Development”

Who is online

Users browsing this forum: No registered users and 1 guest