Inspired by the successful Kinect bounty put out by Adafruit, [gallamine] of the RobotBox community has posted his own $200 $400 bounty for the first person who can hack the scanning LIDAR from Neato Robotic’s XV-11 vacuumbot. This sensor would be particularly useful to any robotic makers out there, because even the full retail price of the vacuum is less than the cost of most standalone LIDAR units, which often run upwards of $1000. The bounty seems to be growing every day, starting out at $200, and doubling thanks to a couple of other interested parties.
Luckily, from what we hear, the sensor was never made to be hack-proof (and perhaps even secretly hack friendly?), seeing as one of the prime developers of the sensor is a member of a certain Home Brew Robotics Club. We love it when companies are nice to hackers, and we hope to see more examples of this in the future. Not sure what the XV-11 is? Be sure to check out the video after the break for info about the vacuum and its scanning LIDAR.
(From a modders standpoint of the product)
I hope a dev can reverse engineer the machine, it looks like a great ‘parts’ machine for the price.
(From a consumer standpoint of the product)
If reverse engineering doesn’t work and you get stuck with it,I hope the 2d mapping system on this is better than my roomba as my piece of crap can’t even dock properly with the charging base.
It seems like the people offering up these bounties don’t realize exactly how much effort and time goes into real hacks of the sort they are requesting.
I praise Adafruit for figuring this out and upping their Kinect bounty to $2000, but for $400 all I’m going to do is put a funny hat on this thing.
For reference, these things are currently selling for $399.00 on Amazon. That means the prize won’t even cover shipping costs.
Correct me if I am wrong:
Did not the 2005 DARPA Grand Challenge…
http://en.wikipedia.org/wiki/DARPA_Grand_Challenge#2005_Grand_Challenge
…put an end to the advantages of an LIDAR system? I mean that is what everyone else was using, including the VERY IMPRESSIVE “Red Team” from Carnegie Mellon University. Instead the (rather small) Stanford Team won. Again, correct me if I’m wrong, but I think they made a very simple assumption that played out very well. They assumed the road to be colored differently then the land, and, accordingly, plotted their course.
I think the problem was that (some?) LIDAR systems rely on spinning mirrors and may also need to be isolated from shock. After taking apart my Roomba, I have real doubts that such a system (spinning mirrors) will survive in the brutal battle ground called “house keeping”**.
**(I have my own theories about why military companies make vacuum cleaners. Simply put, if it can clean your house for a year it certainly can survive a military sortie.)
Here is a link to a PDF that explains a lot about the LDS unit and its principles:
http://www.binary-apotheosis.com/had/RevoLDS_Whitepaper.pdf
macgyver I think you misunderstand what a bounty is. It is not meant to cover your time/costs, it is merely a supplimental rewards to the main one which is that you have the result of the hack for your own use. If you like it is mainly encouragement to share not to do the hack in the first place.
@M4CGYV3R:
I agree, it is a small bounty. But if I paid someone to keep my mythtv boxes running, I could justify buying about a dozen TIVOs.
BTW, I also looked at the Adafruit bounty and thought it would be a tight squeeze to come in at budget. I’d put it at 50/50. And then I know I wouldn’t have finished before [Hector].
(BTW, can someone explain why we put a “@” before people’s names when responding and “[]” around peoples name when embedding them?)
See this paper to get started:
http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4543666
These researchers worked together w/ the guys at Neato so this paper is very relevant.
@st2000:
1. I don’t understand the TiVO / MythTV analogy.
2. The adafruit bounty was nearly 15 times the cost of the device being hacked. Had you won you would make a handsome profit. If you don’t, you’re only out $150 and you have a Kinect.
With a $400 robot, that I would never use otherwise, I can’t justify buying one to MAYBE recuperate only the cost and only use it for hacking.
Funny enough, the technique being used to measure distance with both the Kinect and this LDR is pretty similar.
3. @_____ means you are “talking at someone”. It’s also used in Twitter replies (I believe).
@M4CGYV3R
ok…
1) Mythtv is an open source collage of software that makes a PC work similarly to a TiVo (or ReplayTV). But (I think) it runs rings around TiVo. However, it is a very hands-on effort.
http://www.mythtv.org/
2) Yes, but the saying goes “your time is money”. And as much as I enjoy participating in these discussions, I realize it is actually costing me real money. (Geez, now you took all the fun out of it.)
3) Ah, thanks for the pointers.
This page has a bit more info on the device:
http://www.hizook.com/blog/2009/12/20/ultra-low-cost-laser-rangefinders-actualized-neato-robotics
It seems to work along similar lines to the Sharp IR Distance sensors that are quite common round these parts.
Lidar’s value is not as a Better or Worse sensing tech by comparisons per se. The reason/s for using Lidar are a granularity/blending element adding multimodal capability+ it serving as Communication Redundancy.
Also- IR Lidar+Kinect etc may expand the sensorium of telematics.
Anyone have ideas of building your own version of this? Anyone have good sources of easy to interface linear IR CCD arrays like the Sharp IR has?
Can you replace the IR led in the sharp IR with a infrared laser?
Why would you think it’s secretly been made hack-friendly? They probably just implemented it in a way that’s friendly to the engineers building it. The fact that this is helpful to anyone else coming along and looking at the interface is completely incidental.
What is the point of the bounty? to hack this thing or to create a LIDAR
@razorconcepts, just about everyone will have a CCD hanging around in a gameboy camera, not linear no, but it does have built in edge detection and is simple to use
hey it’s nibbler from ctfxc
why pay a bounty to hack one that’s already made, when someone with the skill could most likely just make one from a couple of old cd-walkmans (possibly?), glue logic and power supply? even if thats the wrong kind of laser, 2 of the right kind couldn’t be that expensive?
What the F are you talking about, $1000 for a LIDAR. Buy any $150 laser rangemeter such as Bosch DLR165 or Leica or Hilti and tap into the PIC or serial bus on it. Stick that shit on a stepper motor and boom, a 2d scanner with 2 mm accuracy over 160 feet.
@razorconcepts check any flatbed scanner in your local landfill for linear ccd and controllers.
Usually they are short onpin count. Like 30 or so.
@ Wartex:
Exactly!
For all intent and purposes, the rangefinder/servo combo is a great option for the price. Done it 6 yrs ago and it worked fine.
Sadly, I never felt it would “expand the sensorium of telematics.” in my case. :P
Somehow I missed this discussion till now.
The bounty has been raised to $800, with RobotShop offering to refund you the cost of the robot if you buy one.
Someone on the SocietyofRobots forum indicated that Neato will make “replacement parts” available soon, which will most likely include the lidar and a lower cost: http://www.societyofrobots.com/robotforum/index.php?topic=12612.0
@Wartex, if you want to document your work and post it to the bounty page, that’d be most welcome! But, I think you’re missing the point of a 1) rugged 2) small 3) low power, portable 4) packaged sensor.
Neato has publically stated that the unit is easy to hack: http://cmusv.na6.acrobat.com/p69585299/
@goeff: How high a sample rate were you able to get with a hacked rangefinder?
@chris
A MEMS scanning mirror would let you get pretty high, and you wouldn’t be wiggling as much mass around.
http://www.robotshop.com/blog/images/Neato-270.jpg
The pads are even labeled… it’s a LVTTL serial port. No idea on data format, though.
@MRE I just took part 2 scanners for my CNC. Might as well try and hack the linear ccd and controller.
@st2000
If anything the DARPA grand challenge demonstrated that the best method of identifying drivable road relied heavily on Lidars. Every team used them to identify curbs, etc, while the most successful teams also used them to identify road paint using some lidar’s ability to deliver both reflectance and range data rather than just range.
The teams that relied on cameras to identify road paint, principally MIT, suffered from computer vision’s most common failing – changing scene illumination (mainly tree shadows falling on the road). Lidars are immune to this as they illuminate the scene themselves with their lasers.
All scanning lidars direct the laser using a rotating mirror but AFAIK they are relatively immune to vibration – instead they are sensitive to direct sunlight. The SICK lidars most teams used are engineered for industrial environments.
A cheap lidar would be great for the home-built robotics community. While some ~$2500 lidars have recently come out from Hokuyo, the 2D SICK lidars used by the DARPA competitors cost ~$5000 each and the 3D lidar each team used from Velodyne costs ~$70k.
Most teams invested almost $100k just in lidar sensors – a number perhaps matched only by that invested in their inertial-aided GPS system. The two in combination were the primary means all of the most successful robots used to perceive their world.
Sparkfun bought a robot and figured out its a 3.3V serial line running at 115200 baud. They haven’t completed the challenge but did a few experiments and provided all of the logic analyzer test files. This way people without the robot can help out too. Take a look below:
http://www.sparkfun.com/news/490
Xevel, on the Trosssen Robotics forum has made some good progress:
http://forums.trossenrobotics.com/showthread.php?t=4470&page=3
Also, the CEO of RoboDynamics is offering $200 to the first person using this sensor for SLAM:
http://robotbox.net/blog/gallamine/open-lidar-project-hack-neato-xv-11-lidar-200-bounty
https://github.com/Xevel/NXV11/wiki
Some info + some code.
Enjoy ^^
How much of bounty will be sufficient to convert this sensor to a Doppler LIDAR for scanning rising air? Scanning distance lets say 0.5 mile around. Scanning data send over BlueTooth to Android or iPhone app showing surrounding area using colors for vertical air velocity?