[size=2][font=Arial]Hello,
We regret that the Sweep LIDAR is dead, however we take the chance to let this LIDAR continue to live - in another form.
We plan to produce DIY kits for a new 360 degree LIDAR based on similar (and some even better) components as the Sweep.
An old prototype of our DIY LIDAR is working since years (grauonline.de/wordpress/?page_id=1233), and the new version is in work. It will have the same LIDAR sensor as the Sweep, a brushless motor with integrated encoder, water-proof housing, WIFI access.
We will inform at Ardumower.de (in the Forum) when the new LIDAR is available.[/font][/size]
[font=Raleway, sans-serif][/font]
[font=Raleway, sans-serif][/font]
[font=Helvetica][size=3]So the first concept ideas continues. We will also use an IMU on the PCB. [/size][/font]
[font=Helvetica][/font]
[font=Helvetica][size=3]
[font=Helvetica][size=3]
[/size][/font]
Sensors to the open source 360 LiDAR.[/size][/font]
For the first version of the OpenLiDAR we will use a slip ring to transfer data between the rotating head and the Main PCB. A good slip ring maybe have a life-span of 45-45 million rotations without considering the impact of other factors like shocks, fast turns i.e.
To calcaultate the life expectancy for the slip ring (which is considered as the weakest part) we will perform a stresstest: at max. rpm rotate permanently and measure the contacts / count the error, i.e. 1-2 weeks and you should be able to estimate the life-span.
Even with a very high guilty slip ring the life-span can be considered as to “weak” for a robot working on a daily basis performing tasks.
if the Lidar turns with let’s say a max. of 20Hz => 4010^6 / 20 = 210^6 seconds =555hrs =23,15 days
and trashed!
To have the ability to increase the life-span and get a smooth rotation for the head we are exploring the option to use shock absorbers between motor => LiDAR mounting (same way as a gimbal camera mount).
To eliminated the need of slip rings we also started working on a concept that uses wireless power transmitting from the bottom housing to the rotating LiDAR head (with help of coils) and the head transmit the data back wirelessly to the main PCB.
Lidar Lite V3 I2C => µC => Transmitter => air => Receiver => ESP8266 i.e.
Looking on diffrent LiDAR´s soultions the LeddarVu seems intresting (which robotshop also sells)
Here I am Simulating 2X Leddar Vu8 in Gazebo. 16 Laser segments. 8@20°/3° and 8@48°/3°. Max Range 121M 100HZ. 360 degree rotation.
Quite nice results running octomap.
Will continue to experiment with this. It would be interesting if one can build (a good 16 segment laser) 3D 360 Deegre LiDAR under 1500k.
The range of 3D LiDAR available today is oh so expensive, starting from 4K ->+ that most people cant afford.
You can find the coil dimensions on the product page under Dimensions!
The SF30 are quite good and work out of the box in a very simple manner. Definitely quite nice, but certainly more expensive and may make an “open LIDAR” a bit out of reach price wise for most DIYers.
As for the LeddarVu stuff, their sensors are really great and their support staff is efficient and knowledgeable, so definitely a good idea to use those if you want “multi-beam” sensors.
We also have some new multi-beam rotational products, but as you said, they are all quite expensive. Our cheapest one is over 3.5 kUSD, which is quite expensive for most DIY-oriented projects.
For interesting solid state “array” LIDAR, you may be interested in the newer RB-Ben, such as RB-Ben-05 and RB-Ben-06 (not as good resolution, but you get a full 320x24 array of points per scan!!!).
Concerning slip rings, they are most certainly a Bad Idea ™ for anything robotics that needs to run continuously to be useful, such as a LIDAR-based sensor.
For example, most newer intelligent vacuums that have rotating LIDARs changed over from slip rings to wireless communication and power internally. These designs are far more durable, of course.
Really cool project and thanks for sharing all those details! We are certain the RobotShop community finds this quite interesting.
Update:
The design of the LiDAR is getting more complete. Using Qi modules and IR data channel to avoid slip rings.
waiting for components to arrive!
Here is a quick and dirty 3D print of the housing (fast print and .6 mm nozzle).
Quick proof of data transfer concept using IR serial communication between two nodemcu’s. Sending AS5048A sensor data wirelessly with very cheap IR diode and Phototransistor. Next up is to try Vishay TSDP series and TSAL6200 that could support up to 9600 bps with a single TSAL6200. Any other suggestions that could support even higher bps?
Something’s up with the attachment - can you check? The concept is very similar to what Robopeak uses in one of their Lidars.
The winner of a hackathon which RobotShop sponsored a few years ago did the same thing (he’s actually in the SESV2 beta test with you) - using IR to transmit / receive larger amounts of data.
I guess it was a .tiff problem.
Never heard of Robopeak before. But It seems alright. The end goal is a 3D LiDAR that is rotating on two axis and the build price should be below 500USD.
For the end product the IR transfer will use **Vishay TFBS4652-TR1 **which can handle [highlight=#ffffff][font=Arial, Helvetica, sans-serif][size=2]SIR 115.2 kbits/s (prototypes parts ordered) and will post updates ones I have something to show [/size][/font][/highlight]
[highlight=#ffffff][font=Arial, Helvetica, sans-serif][/font][/highlight]
[highlight=#ffffff][font=Arial, Helvetica, sans-serif][/font][/highlight]
[highlight=#ffffff][font=Arial, Helvetica, sans-serif][size=2]/RF[/size][/font][/highlight]
So the first prototype is coming together. Changed the Brushless motor from a DJI 3510 to an PM2805 with encoder feedback on the Vertical axis.
The first version will use a LidarLite V3, here we need a really slow vertical movement @ 0.01 hz and 3.0 hz horizontal, so around 100 seconds for one rotation for a dense 3D Cloud (not for realtime 3D, but for mapping it’s ok - like the Leica BLK360)
Later If we change to a faster laser like the SF30-C (which is around 20 times faster) we could have a nice 3D Dense Cloud in around **5 seconds. **This would also mean we are close to the limit on the IrDA transfer. But lets see later on.
All the current parts. Still waiting for Qi modules from China and in the development to make a custom 2A Qi module.
LidarLite V3 fits nice in the housing. Using 6710 2RS / 61710 2RS 50x62x6 mm bearing at top/bottom to reduce load on motor. The motor needs the power to work against the friction of this two bearings.
Very nice to see it taking shape! Those updates are quite nice to read on.
Side note concerning the SF30-C though: while it does have a very high data rate, the datasheet does seem to indicate that when providing measurements at 10+ kHz the resolution is 0.5 m (or 1.0 m @ > 18.3 kHz). You can see this in the user manual, page 8 found on the product page under Useful Links. I figure that may be important since it greatly changes the quality of the point cloud. The resolution of 0.03 m seems to only go up to a refresh rate of ~1.1 kHz, which sounds about the same as the LLV3HP.
Thanks for the feedback guys! I guess the sensor choices will be experiment once we get the motion correct. One could also consider the VU-8. LLV3HP would/could also be a nice choice. But let’s focus on that later on
It’s seems that my topics only attract admins I always think its strange this things don’t attract other users on this forum?
