Outdoors LIDAR question

[Sergei G]

I am looking to buy a LIDAR (<$300) to be used outdoors in bright sunlight with ROS2 Humble.

It will be likely used with Oak-D Lite - for point cloud (obstacle avoidance) and basic object recognition within the 5 meters range.

Any opinions/advice/recommendation? Personal war stories?

Best Regards,

-- Sergei

[Kevin Knoedler]

You can usually get a used Velodyne VLP-16 on ebay for under $300.  They typically have a cut cable, but you can add Ethernet/power connectors and be good to go.  The Velodyne units have good support in ROS2.  A couple of example units:

https://www.ebay.com/itm/156144426764

https://www.ebay.com/itm/135099591000

For a bit more you can get one with a connector and then buy an appropriate adapter.

https://www.ebay.com/itm/404658817303

https://www.ebay.com/itm/234637166040

For yet a bit more you can get one with the interface box.  The interface box is not required.

https://www.ebay.com/itm/355759877021

Obviously purchases on ebay have more risk than buying it new, but the Velodyne unit performs well in direct sunlight and lots of the SLAM algorithms work with it.

Kevin

[Chris Albertson]

Look at the Unitree L1 Lidar, it is just over your budget but I doubt anything is even close to what they sell for $349.

I’ve been looking at their products because of the attention they are getting from low-cost quads and the new humanoid.  The field of view is 360 degrees horizonal by 90 degrees vertically and they claim it works in bright daylight.  They use this Lidar on their robo-dog and it seems to be able to not run into trees.  It completes one 360 degree scan at 11Hz.   I *think?* it uses its interal IMU to correct for platform movement but I’ve not studied the API yet.

one unusual feature I don’t know how I would use is that the Lidar returns not only distance data but signal strength, or ”intensity” data which is actually reflectivity data.   

The best thing about Lidar is that you do not need lots of CPU-power to process it.   But on the other hand, it is hard to argue with Elon Musk’s famous observation that “humans can drive cars using vision only”.  The counter argument is that human drivers kill more than one million people every year.  They set a very low bar for performance.

-- 
You received this message because you are subscribed to the Google Groups "RSSC-List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rssc-list+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rssc-list/DS0PR12MB6439ABE8DB4519FE9F9C9591C9D62%40DS0PR12MB6439.namprd12.prod.outlook.com.

[Sergei G]

Thanks, Chris - this is a strong contender. The budget, for quality goods, is always flexible, when a cute robot asks nicely...

I don't see much info about its operation with ROS2 (Humble is my current platform). Here is a ROS1 driver though: https://github.com/unitreerobotics/point_lio_unilidar

Other GitHub find: https://github.com/tigakub/slam

One concern is how much CPU I will need for the beast. RPi 4 (currently on my bot) seems a bit weak. Especially considering that I have only sensor and "base" nodes there, everything else runs on the "Ground Station" Desktop.

This is the nice robot, Dragger BTW. Being ROSsified/MockTurtled at the moment:

Best Regards,

-- Sergei

---

From: rssc...@googlegroups.com <rssc...@googlegroups.com> on behalf of Chris Albertson <alberts...@gmail.com>
Sent: Wednesday, June 26, 2024 2:24 PM
To: Sergei G <msg...@hotmail.com>
Cc: RSSC-list <rssc...@googlegroups.com>
Subject: Re: [RSSC-List] Outdoors LIDAR question

 

To view this discussion on the web visit https://groups.google.com/d/msgid/rssc-list/D27A30D3-C1C0-4A73-8177-5EE39FC6C6D9%40gmail.com.

[Sergei G]

There's a Ubuntu 20.04/ ROS2 Foxy driver from Unitree: https://github.com/unitreerobotics/unilidar_sdk/tree/main/unitree_lidar_ros2/src/unitree_lidar_ros2

Could be useable.

Best Regards,

-- Sergei

---

From: rssc...@googlegroups.com <rssc...@googlegroups.com> on behalf of Sergei G <msg...@hotmail.com>
Sent: Wednesday, June 26, 2024 3:57 PM
To: Chris Albertson <alberts...@gmail.com>

To view this discussion on the web visit https://groups.google.com/d/msgid/rssc-list/DS0PR12MB6439FB858D8FD874989E2F52C9D62%40DS0PR12MB6439.namprd12.prod.outlook.com.

[Sergei Grichine]

Thanks, Kevin - this is an interesting option. Ethernet connection is fast, and cable soldering doesn't scare me. Unknown condition is a bit of a downside though. Newer designs could be also more capable.

--
You received this message because you are subscribed to the Google Groups "RSSC-List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rssc-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/rssc-list/CAPH8A1CinDfkTNceY_%3D8-1_gBc7G9NRXAYehzjm1G%3DpH90UpvQ%40mail.gmail.com.

--

Best Regards,

-- Sergei

[Chris Albertson]

I would not worry about publishing the points on ROS2, porting a ROS1 driver is easy, as the hard part is always interfacing with the hardware and that would not change.

The MUCH harder part is doing something interesting with the published point cloud.  you would not be doing the classic SLAM stuff so you’d be on your own for that.

As for computation, I think it is all on the subscribber's side.   Publishing can be nearly free on ROS2 if you can set it up so “zero copy” can be used.  If the publisher and subscriber(s) are in the same Linux process then the data can remain in a RAM buffer and never move.   

That said, once you start doing things like object recognition and keeping a database of objects, it might be time to add a second Pi, Maybe a Pi5.  That robot is large enough that there is space to add several more computers.