Welcome to RobotBox!

RobotBox is a community for robot builders to show off their projects. Add yours today.

Features of RobotBox

  • Showcase projects
  • Make new friends
  • Rate other's robots

Note: Facebook Connect temporarily disabled. Go here to login without it.

Neato Robotics XV-11's Lidar Hacked!

gallamine's picture
Posted Fri, 11/26/2010 - 3:04pm by gallamine

15 The Open Lidar Project - Hacked!

Hash79's setup for hacking the Neato Robotics XV-11's lidar

Ten days after posting the bounty for hacking the Neato Robotics XV-11's lidar, I think we have a winner! Hash79 (blog)  has posted a video of his hacked robot outputting lidar data. You can see it below. Hash79's work is based on code from Xevel, but it turns out that Hash79 has a XV-11 with a firmware version 2.4, which is significantly different than the 2.1 firmware output data provided by Sparkfun.

If you'd like to continue participating in the hacking process, there's a very active thread going on on the Trossen Robotics forum where most of the work was performed. Be sure to congratulate Hash79 on his hard work! The wiki page for Xevel's code has been updated with all the known information on the 2.1 and 2.4 firmware output data, and is a great place to start if you'd like to contribute. Hash79 also has a good explanation of the output data on his blog.

The box used to give distance data for the Neat Robotics XV-11 lidar

Hash79 writes,

I had a blast working on this, nice finding new blogs to follow along
the way as well...  Looking forward to the next challenge!  :)

For now, if you want a sensor like this, you'll need to purchase one of the XV-11 robots (from RobotShop or Amazon), but hopefully Neato will hear our pleas and make replacement units available at a lower price.

Comments

hash79's picture

Simple motor control circuit

by hash79 - 12/08/2010 - 3:17am

Simple motor control circuit built using parts ALL found at Radio Shack.

http://random-workshop.blogspot.com/2010/12/simple-circuit-to-control-lidar-unit.html

-Hash

chrlsrchrdsn's picture

Hash79, What is the sweep

by chrlsrchrdsn - 12/02/2010 - 10:34am

Hash79,


What is the sweep area of the NEATO Lidar?


Rick

hash79's picture

Rick, Supposedly full 360

by hash79 - 12/02/2010 - 7:07pm

Rick,

Supposedly full 360 degrees out to 6 meters from the sensor.  I have not verified this though, that is information from their white paper before this production level LIDAR unit was designed/built.

-Hash

chrlsrchrdsn's picture

gallamine, It may be worth is

by chrlsrchrdsn - 11/30/2010 - 6:34pm

gallamine,


It may be worth is for specific people at specific times.  $400 is still about .337 of a low end SICK and its got better range and the mapping is more reliable.  BUT yes, I am in that group that thinks $400 is still way too much, though Hash makes a great point on using the vacuum parts for other projects.


BUT for the love of being able to get my hands on a reliable slip ring and Linear CMOS that are both available if a distributor would make them available, we could generate better, longer range, more flexible designs and execute them cheaply. 


Hash, the two most costly parts of that device you hacked was the slip ring and the sensor.  With todays available sensors at available distributor costs + profit, we really should be able to make a basic sensor for about $45-$75 depending how fancy you want to make it.


What would be a really interesting hack would be to nod it and get a limited 3D view of what is in front of you.

hash79's picture

chrlsrchrdsn, We are

by hash79 - 12/01/2010 - 5:47am

chrlsrchrdsn,

We are discussing the idea of selectively scanning an area on the Trossen Robotics forum right now, although it is high res selective 2D scanning. 

http://forums.trossenrobotics.com/showthread.php?t=4470

Take a look at the detailed tear down pics of the LIDAR unit I just posted

http://random-workshop.blogspot.com/2010/12/lidar-unit-disassembled-further.html

I believe it would be trivial to spoof the encoder data and cause the LIDAR unit to believe it is spinning and scan.  Placing the rotating turret on a stepper motor with our own encoder would allow us to control position and then output our own data stream selectively scanning areas.  Maybe keep the unit stationary and use a couple mirrors to scan a 3D scene.  Haven't thought that far into it yet.

-Hash

chrlsrchrdsn's picture

Hash79, Good idea about the

by chrlsrchrdsn - 11/30/2010 - 6:27pm

Hash79,


Good idea about the parts.  They are probably cheaper as a purchased vacuum than separately. 

chrlsrchrdsn's picture

I guess I am a little

by chrlsrchrdsn - 11/30/2010 - 12:09am

I guess I am a little confused.  The idea is to buy a neato vacuum for $400 for a sensor that the foundation paper says can be made for $30 (I think closer to $45) and pressure Neato into what? Why would they build a cheaper one if they can get everyone to buy $400 little robot?


I would be much happier if I could find someone to sell me a 600 rpm, 6 circuit slip ring capsule available in qty 100 for $12 and get a distributer to sell me a 1024x48 linear CMOS sensor available in qty 1000 for $16 (that has global shutter). I have the .9 degree stepper motor and green laser that will give me about 10-12m assuming I don't care if minimum range is over .6m.


I propose an open project that comes up with the standard triangulation software that will let the user set up the min/max range of the LIDAR, along with laser pulse duration and timing.  Set up a kit that can be vertical or horizontal in configuration and add 3D.  This is doable with sensors and slip rings, but try finding them for sale to individuals.


Then Neato can change their firmware anytime they want.


Hash79, I am sure you worked hard and it would be pretty neat to have that sensor on a robot.


Trossen, spend some time on 6 or 12 circuit slip rings capsules, preferably with a through-bore in the 3-8mm range that can do 600 rpm.  OR look into those Linear CMOS sensors! (and don't forget they need global shutter...) Then watch the LIDAR light up the world.

hash79's picture

I'm all for an open hardware

by hash79 - 11/30/2010 - 1:53am

I'm all for an open hardware design.  Im sure we could do it for sub $150 in parts in low quantity.  We could also use RF to transfer data like they initially did to avoid costly/hard to get slip rings.

But if you look at the XV-11 it also contains a lot of good parts for someone making their own robot so $400 isn't too bad for all the sensors, drive wheels/motors with encoders, etc.

-Hash

arthuroforange's picture

Well what are we waiting for

by arthuroforange - 08/31/2011 - 7:10pm

Well what are we waiting for ? the paper talks about a $30 lidar unit. We can surely do something similar atleast in the beginning. If one had the resources one would certainly do it. They talk about this REVO LDS (lidar) but apparently it doesnt exist and no one is making something similar which is unfortunate. 

gallamine's picture

Are you putting money on it?

by gallamine - 11/30/2010 - 12:10am

Are you putting money on it? :)

If you go back and read the original bounty post, you'll see there's some context posted. Even if Neato doesn't make cheaper units available, the sensor is still worth the $400 you have to spend to get it.

hash79's picture

Thanks Xevel, and great job

by hash79 - 11/29/2010 - 1:23pm

Thanks Xevel, and great job figuring out the checksum for the data stream!

-Hash

Xevel's picture

Congrats Hash!

by Xevel - 11/29/2010 - 8:39am

Congrats Hash!