3D LIDAR Scanner (new version)
HTML-код
- Опубликовано: 23 окт 2020
- This 3D Lidar improves on my previous Lidar project with a better mechanical design to achieve much faster scanning rate and higher resolution. It relies on a mirror to scan the beam around the room, and has not moving electronic part. This video describes the design, construction and usage of this sensor.
Original article: charleslabs.fr/en/project-3D+L...
GitHub repo: github.com/CGrassin/3d_lidar
CC-BY Charles Grassin
Made with open-source software: Kdenlive, Processing, Arduino, Inkscape on Ubuntu. 
Наука
This is brilliant, well done. This is on my project list now. Many Thanks Charles.
Thanks for sharing your project. I like it a lot. In the past I did an scanner with two servos, but yours is better. You've expended lot of time for preparing and sharing details of the design, so thanks again. Juan.
I am working on a similar project that involves a 2-DOF system with yaw and pitch capabilities, along with a ToF LiDAR. After scanning, I obtain data that includes distance, yaw angle, and pitch angle. I use MATLAB to convert these into Cartesian coordinates and to create a 3D view. However, some lines that should be straight appear curved. Do you know why this might be? I intend to use this setup for 3D reconstruction of fluid objects such as sand and gravel, in order to calculate their volume
Totally awesome! I'm definetly going to try this one out. Good work!
Awesome project, and pretty good article.
This is killer! I gotta build something like this.
The mechanism impresses me
Interesting idea with the mirror !
good for those who don't have the expense to afford a commercial product, btw, first surface mirrors are preferred as the speed of light in glass is much lower than in air
simple, cheap, scalable quality, perfect
thats awesome good job mate
Thanks :)
Great project! Thank you!
very nice, would love to see it hooked up to SLAM
Man, you got my attention with this one! Thanks for sharing!
I also have some improvements ideas like the other colleagues:
1. ball bearing for the rotating mirror it's mandatory
2. belt drive for the rotating mirror should be a better choice in terms of motion fluidity. Gears if not printed correctly, will induce vibrations into the mirror.
3. mounting the linkages for the mirror angle on at least 3 vertical rounded shafts (like those used on 3D printers axis), around the rotating assembly, together with another ball bearing, to ensure a parallelism between the base of the whole lidar system and the angle control ring for the mirror.
4. a DC motor with encoder (you can find in old paper printers) should be a cheap option to rotate the mirror assembly in a fast and controlled manner.
5.
Thank you for your comment! I agree with all of your proposed improvements. However, it would make it more complex to replicate. I might do a version similar to what you described in the future, but also maybe an improved fully 3D-printed version.
@@CharlesGrassin your design still best on youtube or other platform, its like nobody doing this. do you still progress on project or it will be last version we watched.
Hi, I am a bit late but is it possible to use stepper motor instead of the DC motor?
Thank you for posting!
when someone talks about 3D LiDAR scan. I think it's best to distinguish between: inside-out scan and outside-in scan
they are different approaches.
the inside-out like the one in the video, has a problematic of many blind spots, to fix that, you don't need a 360 rotation method, but the entire scanning station should be mobile.
the outside-in scan on the other hand, can use a 360 method and it will be actually perfect.
Well done!!
Awesome LIDAR project. I liked how you solved the motion of the mirror.
A few tweaks and it's perfect.
1. As others already suggested, use ball bearings on the axis.
2. Instead of using a gear link from the motor to the rotating assembly, use a timing belt. It would lessen the vibrations due to slack in the gears.
Note also that you do not get the same length from all the lidar angles due to the sensor's construction.
Thank you for your comment! In this case, I wanted a fully 3D-printed construction. But I agree with your suggestions. I will probably do an improved version in the future with a timing belt, a bearing and rods for the pitch axis.
I think it would be beneficial to instead use BLDC's, you can set the angle directly (sin pwm with mosfets), it would need some balancing work though, but the TF Luna can be set to update 250 times while stock is 100, and the TFmini-S up to 1kHz (1000 dots a sec), maybe more. that'll allow much faster scanning and movement.
Super chouette comme idée :0
nice work
Excelent!
I am just replication your project an i have all the printing parts fit together. But now i wonder why you didn`t made the rotor higher, so that the mirror can move more down. So that you have a wider scanning area. I want to do a scan of a room, and therefor i also need to scan the ceiling. I would be really happy if you could answer :) Thanks in advance.
Nice, thanks. Next: Let's put the LIDAR as the eyes of a robot, simulating a saccades. Then sensor fusion.. and create a representation of the visual environment... 😅
So is a standard car mounted lidar spinning the sensor because that keeps the whole thing smaller?
Cool idea. How does it manages to keep stock range value of TF luna when its reflects on mirror and back?
Thanks!
I don't have actual numbers, but it doesn't really affect the range measurement in my experience. The additional travel path is only a few centimeters and the reflectivity of a standard mirror in the wavelength of this Lidar is well above 90%.
Hello, I am getting an error in the processing part of the project, it does not read the lidar. When I test the lidar it works. Would you help me?
what is the maximum detection range of this lidar?
would there be an affordable way to get 30~45 meters of range?
The TF-Luna only does 12 m max range. The Gamin LIDAR-Lite does 40 m range, but it is much bigger and more expensive. I used it in a previous scanning 3D Lidar project: charleslabs.fr/en/project-3D+Lidar+Scanner
Brilliant
Amazing
does this method reduce the sampling frequency compared to a stationary lidar ? I'm interested to know it's drawbacks
Hi, I think it reduces range and number of point clouds due to accounted losses in the system.
Could you discuss the math behind the scanning a little bit more? I would be interested in hearing how that is achieved. Merci beaucoup.
How many degrees it can get? From 0 until 45? Because its with mirror reflection method... i also research with 2 mirror reflection like in panasonic lidar style, so for mirror 1 is fixed in 45° for transfer the laser into mirror 2 is rotating by pulley and also set the degree with micro servo there to make field of view. What do you think? Im still get arround 20-40° effective to capture the room bcause 0-20 still detect the body of the lidar and 20-40 is just about seeing mostly the road/bottom side. The lidar couldnt see the object on the forward-high side. (45° above)
Relative to the ground, it's about -25° to 20°. This range is determined by the length of the link rods and can be changed depending on the need (of course, the code also has to be changed to give the correct angle).
There are many interesting configurations for the mirrors. Panasonic 's Lidar is indeed one of them!
Oh okee.thank you..nice sharing. -25 to 20 is for swing that 2 servos yaa... if its directly mount to control the mirror its like 0-45° if relative to the ground until the mirror can transfer the laser straight horizontal. What type do you use for this 2 servos? Mg90s 90 degree or 180 degree? With mapping? This angle setting not same ya For the right and the left servo? If the 2 rods controlling the mirror angle together like this. Its probably like the right servo set -15° and the left servo set 15° to get 15° actual in the mirror angle.
This is often how a sea based radar works, usually thy got dual levels on them
Should include a lidar room mapping as proof of concept.
cool
yoooo a swashplate lol damn nice
You could use only one stepper motor and one potentiometer
If you used a CIS sensor and a laser with a lens to transpose the beam into a line it would be far simpler and effective.
why did you use mirror ? i mean you can use Lidar sensor directly
The mirror moves the laser beam on two axes without having to move the lidar sensor itself. Because a mirror is lighter and has no wires, the scan can be much faster and more reliable.
Add this to the roomba and you have a house interior scaner
Es-tu québécois?
Non : Paris, France 🙂
Je me demandais la même affaire lol
but where to keep the object
Pas mal mdr