Starting in 2009, I noticed the need for LiDAR. Ground and aerial bots need good data from their physical surroundings to "prime the pump" of navigation methodologies like SLAM. Seeing others and myself having to cope with this "sensory-bottle-neck" was frustrating. We'd have this great microprocessor system, advanced program code, and really capable bot, but without LiDAR and SLAM, we couldn't trust the machine to fly around autonomously.
So, wanting to participate in the 2011 International Aerial Robotics Competition (IARC), I started looking at sensory solutions. Some of the teams could afford the Hokuyo LiDARs. These were 1200 to 6500 US$, and the modern ones are USB only. Other teams used home made solutions like scanned ultrasound, crossed laser beams and so on. But until I saw the Neato XV-11 hack challenge, I wasn't sure where to go next. Fortunately, seeing autonomous flight attempts from 2011 to 2013 in Grand Forks North Dakota, I became more and more convinced that sensory and telemetry systems are the key to getting our UAV's to be more than just expensive toys.
At CSU Chico, my Electronic Engineering and Computer Science peers and I followed the XV-11 LiDAR and Microsoft Kinect reverse engineering bounty. Including some reverse engineering and operating of the structured light optical train ourselves (really cool by the way!). But seeing the CMU cam 4 and the XV-11 LiDAR really brought my mind to focus. A spinning LiDAR is unavoidable for the hobbyist, high school through college students, and experts alike.
The lack of - and need for - an affordable open source LiDAR has become an intolerable obstacle, and this must change.
I want this device to follow the spirit of the CMU cams and Pixy 5 optical flow cameras. This LiDAR should be easy to attach to Arduino's, PICs, STM32s and Basic Stamps. The interface and coding should be simple, and will be thoroughly documented in a PDF. Example code and sample setups will be included. This LiDAR is meant for High School teachers and students all the way up to college professors and students. My goal is for more LiDARs in the market place, at lower cost.
Like the XV-11, it will have a mechanically scanned sensor. This method has been used in industry for decades for radar and other sensors. The sensor is from the PulsedLight3D LLC group and this device, the "LidarLite" version, is already proven and certified for human eye safety. The turret housing the LED and detector will spin 3 times per second, and the data will be returned serially through a slip ring commutator. The motor will be held at a constant RPM with a closed loop speed regulator. each of the 100 pulses per second distance measurements, will be indexed to the degrees of a compass, as in other common devices.
Mechanically scanned, (spins).
4 degree resolution.
No blind spots, full circle FOV.
Small, cheap, light weight.
Easier to use than XV-11 module.
SPI / I2C serial comms, and low power input.
Range: 0 to 30 meters at 30% target reflection.
100 measurements per second.
LiDAR-Lite, (laser version) with an OEM agreement 60 $, at 200 units.
4 pin commutator, 15-17 $, depending on quantity.
Motors, bearings, belts about 4 $
Circuit Board, 1.25 $, components 12 $
Case and miscellaneous, about 5$
There are labor and equipment costs as well. Ill keep these down by doing things with friends and family, but ill still need to pay for the help.
Work So Far:
Ive put in more than 2,000 hours and 4,000 dollars of my own time and money researching, building and learning. I wouldn't be asking others to do the same if I didn't believe in the critical importance of this technology. So, I currently have the rotating test device, which has verified the geometry and signal timings. next up, there will be code written for a STM32L1 discovery board to interface with an actual lidar lite.
Seen in this clip, the rotating turret and oscilloscope. Ill have the STM32L1 attached soon, to read the IR break beam encoder.
For this project, its important for potential backers to know that I have all the necessary instruments, and knowledge to use them. Oscilloscopes, Logic analyzers, Volt meters, Logic programming, soldering, CNC milling, Circuit board etching tank and so on. So this isn't my first rodeo :) And feel free to look at my previous work in UAV's and High Voltage projects. Further, Im using our purchasing power in quantity to press the price down. So we the end users will get greater capability then just the straight line LiDAR-Lite itself.
Links To My Previous Work:
Academic Bicopter Using Castle ESC's [ It Flies! ]
The risks are mostly related to economy of scale and purchasing power. Meaning I need some fixed, one time expenses like research and development of the optical train. After that I can run the devices out quickly, with more backers this becomes more favorable.
I have costs figured out, with formal quotes for quantities of the components, including alternates if need be.
There are others offering solutions, first RoboPeak, which sells for 399 US$. And then a company that re-sells the XV-11 module with a yellow case online, that I cant find at the moment.