Low Cost Lidar crowdfunding suggestions

[Maeth]

Hey all,

It is really important for us that you take a few seconds to fill out the poll! ^

We wanted to make a small, low cost and easy to use LIDAR (laser based radar in our case)
We are now working on commercializing it, and selling kits while maintaining everything open source! (A la Arduino)

My question is:
Would you buy a ~$100 LIDAR that would provide you with an easy to use 2D map of a room, and also images and 9 axis IMU data?

We are also currently considering using a line laser, and hence getting 3D map data.

Check out our early prototype here:


Details are available here:
http://www.eng.uwaterloo.ca/~ibaranov
at the first link, or directly here:
http://www.eng.uwaterloo.ca/~ibaranov/files/Group_50_FYDP.zip

We are planing on doing a Kickstarter for people to get their hands on kits.
We welcome feedback and suggestions for the business/kickstarter side of things.

I can gladly post our costs estimates and so on for people to pick apart


EDIT:
We will be providing examples how to use our unit, and some code, to produce full accurate SLAM such as the one below:

[Rasz]

Are you  connected to one of Darpa challenge teams? I  seem to remember one of the teams announcing development of low cost lidar for hobby market a year or maybe two ago.

[Maeth]

Nope, no connection whatsoever.

I had not even heard of them, care to provide a link?

[Corporate666]

I would be interested but it would completely depend how easy to use it is and how accurate it is. 

[Maeth]

Good question.

Accuracy:
We had some software issues with the prototype seen above. So we were only able to get ~0.5cm precision, 1cm accuracy over 2m radius.
However, this should be improved to 0.1cm precision, 0.4cm accuracy over 8m in production version.

Ease of use:
We will have a I2C connection, and USB connection.
The I2C will have support examples for arduino, raspberry PI, and most other popular platforms to get you going.
The I2C connection will be able to output distance and angle data, along with IMU data. This can then very easily be assembled into a map. (as above in the video)

USB will allow for more complex processing, such as depth mapping overlay on images, and creation of 3D maps with image data.
There will also be examples provided in a few languages (Python, and processing current ideas, but not really studied yet)

Lastly, the USB connection can also implement a ROS message handler, if people want one. (not sure of demand for this)

[Smokey]

These probably aren't the guys you are thinking of, but they are still totally awesome and Darpa Grand Challenge related.  I love this story.
http://velodynelidar.com/lidar/hdlproducts/hdl64e.aspx

They entered the Darpa Grand Challenge the first and second years it was held.  Almost everyone else was using SICK Optic fixed lidar units but they created their own rotating lidar system from scratch.  The next year they didn't actually enter the race, but what they did was sell a commercialized version of their lidar from the year before to the other contestants.  It was hugely successful with many of the teams that finished the race using their sensor. 
It's actually a multi-laser rotating time of flight laser system with a price tag of around 75K.  You can see one prominently featured on top of the Google driver-less car.
http://en.wikipedia.org/wiki/Google_driverless_car



As for a line scanner for under $100, how does your hardware compare to the Neato XV-11 sensor?
http://xv11hacking.wikispaces.com/

[Maeth]

Ya, the velodyne stuff is awesome.
Wicked expensive, but very cool. I actually got to play with it during one of my co-ops.
Seeing an office rendered in 64 strokes of light is interesting.

The Neato was actually the inspiration for the project.
As Neato does not want to release a stand alone sensor, we decided to make our own.

We use slightly different tech (cellphone camera while they use line sensor).
And we include an IMU.

Ours is admittedly slower (~3 seconds for 1degree scan of 360 degrees), and possibly slightly less accurate.
But we think the added ease of use, extra features, and lower cost would offset the reduction in speed.

Fundamentally, the 120FPS rate of the camera limits our maximum speed.

[Psi]

To be useful to me it would have to be able to scan the entire room within 5 seconds and be able to scan a space of 50 meters length/wide/height.

(And yes, i know that's several orders of magnitude better than it will ever do.  Just saying, if it did i would have a use for it, It would go onto my RC plane for a crash avoidance system).

[ptricks]

(yes i know it wont be able to do anything even remotely near that.  Just saying if it did i would have a use for it. It would go onto my RC plane for crash avoidance code).

I don't do RC planes, but I am interested in it and so attended a local meeting when they were out flying. One person had a plane that had navigation and he said it was done with ultrasound, I don't know the details but if you are interested I'll try to get more info next time.

[Maeth]

The problem with ultrasound is that it is very wide.
Even the very best ultrasound distance sensors have a beam angle of 2 degrees.

This makes for very low resolution mapping.

[Maeth]

We are planing to launch a Kickstarter campaign once we have a production ready design and figured out all our costs to the cent.
I want to avoid the trap of other hardware projects that go over budget.

We should be launching in September, and it is great to see people supporting this project!

[rivercaver]

This sounds like perfect fit for cave mapping.  Will post this on the cavers forum.

[Maeth]

Awesome! Thank you so much!

Would you mind linking here once you do?

We really need all the feedback we can get.

[daedalus]

this type of laser pointer on a stick scanner has been demonstrated many times, there is even a commercial module that uses this trick, the lidar part out of neato robotics vacuum cleaner. The scanning speed really sucks, even compared to a scrap sick safety lidar from ten years back (which can be found on ebay for around $100) and unfortunately there isn't much you can do to up the speed, as fps of the camera is a limiting factor. I personally would love for there to be affordable lidars, as its a limiting factor in robotics research at the moment.

What I think you should be doing is looking at ToF based lidar approaches. There are TDC chips available off the shelf now, and this approach has the ability to scale to very high point counts, TDCs capable of doing enough points to play in velodynes sandpit are only $100 a chip in singles. One really awesome approach would be to couple a ToF lidar system with a mems steerable mirror, which would provide an almost solid-state scanner capable of providing a 2d pointcloud of the scene. This was attempted in a European research project between (IIRC) Sick and Fraunhoffer a couple of years back, with the aim of producing a cheap lidar for car manufacturers. The name of the project escapes me right now.

I think my main point is I would rather give much more money to a project that produced something that has the scope to be useful in the real world, and I just don't think you can shave enough off the manufacturing cost of neatos solution to make a viable alternative, whilst there is a massive amount of money you could shave off velodyne style devices and still be profitable.

[Maeth]

I agree, this has been done before by Neato.
However, there are a few key points:
1) Neato refuses to sell the lidar unit by itself, rendering the cost of such a unit the cost of the entire vacuum cleaner (~$300). We are 1/3 of that with far more capabilities!
2) Neato uses a line sensor, while we use a camera. This allows us to do 3D scanning, in addition to providing video and still images to the user.

Would you be able to provide a link to ebay showing those ~$100 sick safety lidar? That would be greatly helpful!

The problem with TOF is a few:
1) There are tons of giant companies doing TOF sensing (Microsoft Kinect to name one). We cannot compete in Tech or cost with them
2) Manufacturer of TOFs (since they are a new tech) refuse to sign NDAs or sell to a small startup like us
3) A more practical concern, we do not have the time or money to completely change everything about our system to a design that large companies still struggle with. 

Why would you believe the sensor is not useful in the real world? We are not aiming towards high speed industry, we are going towards low cost R&D and hobby markets. Are those not useful markets?

In summary, I should probably change the emphasis of the device. Not just a LIDAR, but also IMU (with proper filtering done for you), Camera and 3D line scanner.

[Pippy]

Microsoft Kinect doesn't do TOF, they use speckle pattern matching. The laser creates a static pseudo random like pattern across the cameras field of view. The camera image (a standard 60Hz NIR camera) is past onto the DSP chip to extract the 3D map.

[Maeth]

You are entirely right, if we are talking about the old kinect.

The new one uses TOF, as this is both cheaper and more easily mass produced

[Pippy]

The new one uses TOF, as this is both cheaper and more easily mass produced

Never knew that. That's a nice bit of technology!

Wonder how it's done exactly.

[Maeth]

If you are interested, look here:

http://www.ti.com/ww/en/analog/3dtof/index.shtml?DCMP=analog_signalchain_mr&HQS=3dtof-pr

[daedalus]

Here is a PLS101 that went for 97gbp :
http://www.ebay.co.uk/itm/SICK-PLS101-312-Proximity-Laser-Safety-Scanner-/271248210031?pt=LH_DefaultDomain_0&hash=item3f27a7386f

I have seen these go for as low as 50gbp before.

As for time of flight, specifically pulsed laser ToF, the major complexity is the TDC circuit, and these are available as integrated ics, without any NDA, and have been for several years. I have even seen university student projects using them for time of flight experiments. ACAM messatronic was the manufacturer IIRC. You could put together a working ToF pulsed lidar from off the shelf pcbs (laser driver, diode, pin photodiode, transimpedance amp, TDC) its not exactly rocket science.

Having a camera isn't really an advantage for a 2D lidar, a line sensor in the same design would usually provide higher framerate, therefore higher scanning speed. Also it requires less processing overhead. This is probably the reason for neatos choice. Since the neato module has been reverse engineered, and the sensor in it is capable of running much faster then neatos design, I don't see how your approach could beat a neato clone in 2d lidar performance. Granted you have the advantage of being able to capture pictures of the scene, and the possibility of 2d.

I agree, this has been done before by Neato.
However, there are a few key points:
1) Neato refuses to sell the lidar unit by itself, rendering the cost of such a unit the cost of the entire vacuum cleaner (~$300). We are 1/3 of that with far more capabilities!
2) Neato uses a line sensor, while we use a camera. This allows us to do 3D scanning, in addition to providing video and still images to the user.

Would you be able to provide a link to ebay showing those ~$100 sick safety lidar? That would be greatly helpful!

The problem with TOF is a few:
1) There are tons of giant companies doing TOF sensing (Microsoft Kinect to name one). We cannot compete in Tech or cost with them
2) Manufacturer of TOFs (since they are a new tech) refuse to sign NDAs or sell to a small startup like us
3) A more practical concern, we do not have the time or money to completely change everything about our system to a design that large companies still struggle with. 

Why would you believe the sensor is not useful in the real world? We are not aiming towards high speed industry, we are going towards low cost R&D and hobby markets. Are those not useful markets?

In summary, I should probably change the emphasis of the device. Not just a LIDAR, but also IMU (with proper filtering done for you), Camera and 3D line scanner.

Edit: first link was to broken unit, updated

[Maeth]

I hate to be pedantic, but you linked to a "as is" marked "Bad" Lidar.
Actually working models of that one cost ~$2000 or more.

That is entirely surprising, as any manufacturer I actually talked with refused to release applications data, even when we were a university team.

I don't quite understand what you mean by your last paragraph there...
That a line sensor is faster? Undoubtedly. It is also much more expensive to do (once you consider that optics must be custom made for it)
The neato solution works because they make 1000s of em, which we cannot do. Line sensors are not even find-able in single quantities anywhere.
The processing overhead is not really of any importance, since it is all done on our chip. The user just gets data.
Is there a neato clone I am competing against? I haven't seen one.

EDIT:
Yes, ok there are working SICK lidars in that range.
Still, you are comparing a used ebay device with a brand new one.
Also, the device is heavier, larger, consumes more power, and harder to interface with than ours, on top of not having any IMU whatsoever.
Not really a fair comparison...

[daedalus]

I struggle to see why you cant just mount the line sensor with the same lens you are currently using with your webcam. I don't have a neato to hand to look at it, but if memory serves its just using a threaded plastic lens module out of a webcam, which are available for next to nothing at alibaba.

The neato sensor is a panovision one, alrad carry them in the uk, and although ive never bought that part, i have bought singles of parts from alrad before.

[Maeth]

Our camera is a cellphone style camera, integrated module and optics. It is very small, much smaller than the line sensors.

Unfortunately no, that is not the case. Even if they are using a standard casing (which I doubt, see link) they have to use a different type of lens to focus light properly across a wider sensor than normal, along with aberration issues with linear sensors. You can ignore them of course, but I am doubtful if this would provide accuracy.

See image here: https://www.sparkfun.com/images/newsimages/XV11Teardown/TheGoodStuffUnmasked.jpg

From my quotes when we were just starting this project a year and half ago, they were very expensive in low quantities (<1000). ~$40 to $20 each.
Our camera module is 4X to 2X cheaper than that, on top of being easier to service and replace.

Regardless, we are not trying to recreate the Neato solution. I actually only learned of the Neato one after we had already decided on this as a project.
We are trying to provide more functionality for less cost, at less scanning speed.

[Pippy]

btw Maeth, I did a similar distance measuring setup. Although I'm using a monochrome camera (to get the highest pixel resolution) with a red 650nM laser and 650nM filter on the camera. The filter and monochrome camera combination makes a big difference.

A quick test I did 6 months ago using a colour camera and the monochrome camera side-by-side ..



The monochrome CMOS sensor itself you can buy on ebay for about $3 each the last time I looked. It has ROI (so can sample just a strip of pixels across the sensor etc instead of the whole sensor pixel array) and everything done on the sensor, you just set it up and read out the pixel data on the bus pins. All ADC (10-bit) etc is on the sensor. Very easy to use.

[Maeth]

Hey that is really cool!

Would be willing to send a link to the filter you are using?

We searched for a long time to find a camera module that fit our needs, and it is color, so we can't really change that....