The purpose of this repository is to organize resources for implementation of a laser fence that targets insects, including the mosquito and spotted lanternfly. A research question is to determine if work done on laser fences that target mosquitos in Africa could be adapted to spotted lanternfly in Pennsylvania. Note that the original mosquito use case was popularized in 2010 and described as a failure in 2016 (Schiller, 2016).
Characteristic | Value | Source |
---|---|---|
Average Flying Altitude | 2 meters | Kare, 2010 |
Supply | ∞ | Kare, 2010 |
Size | < 2 centimeters | Kare, 2010 |
Speed | < 1 meter per second | Kare, 2010 |
Lethal Energy Dose Req’d | > 50 mJ in 3 ms | Kare, 2010 |
“As early as July, adults can be seen, and they mate and lay eggs from late September through the onset of winter.” (Wikipedia 2020) Therefore, a system that target adults should be placed into service from July through the onset of winter.
Characteristic | Value | Source |
---|---|---|
Average Flying Altitude | TODO | |
Supply | “killem killem all” | https://youtu.be/1g82I7w3ENU |
Size | TODO | |
Speed | TODO |
TODO - Translate size to requirements for identification
TODO - Translate speed to requirements for response time
Currently working on the Laser Turret solution.
- NVIDIA Jetson Nano
- Master control system and image processing for target identification. $99.00 at Micro Center
- Erector by Meccano - Super Construction Set
- Turret assembly and component positioning. $199.00 (v1) at Amazon
- Leopard Imaging 145 FOV Camera
- Note that this camera has been discontinued. It was originally included in the older SparkFun JetBot AI Kit.
- Edimax 2-in-1 WiFi and Bluetooth 4.0 Adapter
- Connectivity to the NVIDIA Jetson Nano. $19.95 at SparkFun.
- Laser
- TODO
- Galvo
- TODO
- Power
- TODO
- Cabinet
- TODO waterproofing enclosure with windows for sensors and lasers. The enclosure could have a mask/shutter to prevent firing beams ouside of the desired area. Note that many lasers may have temperature limits beyond typical ambient summer temperature in the target area of operations.
Component | Price |
---|---|
NVIDIA Jetson Nano | 99.00 |
Edimax 2-in-1 WiFi and Bluetooth 4.0 Adapter | 19.95 |
Total | 118.95 |
- Fixed Platform
- Camera and laser are fixed in place; lasers are aimed using galvos.
- Moving Platform
- Camera and laser are mounted to a vehicle; lasers are aimed by driving the vehicle using servo motors.
Callibrate laser beam angles to video positions.
This suffers from parallax error. Kare mentions this but doesn’t address it. There are solutions to create depth maps using a single camera (MonoSLAM, LSD-SLAM), but they require a moving camera in a static environment. We could estimate depth based on camera focus if we could get that information from the camera and it was able to focus on the insect, but this is not likely.
One option is to accept that we don’t have depth, and the insect could be anywhere from the nearest point the camera can see to its farthest point, and straffe the laser from one to the other. The drawback is that this is more laser energy that will strike the background.
Use two cameras to locate objects in real world coordinates (ref).
This doesn’t look precise enough to find the depth of an insect.
- [X] Jetson Nano environment setup
- [ ] Raspberry pi environment setup
- [ ] Detect a target in a single frame
- [ ] spotted lanternfly
- [ ] mosquito
- [ ] Track a target through multiple frames
- [ ] spotted lanternfly
- [ ] mosquito
- [ ] Test laser functions using a low powered test laser
- [ ] Can engage / disengage test laser
- [ ] Can steer beam on target
- [ ] Can limit beam to bounding box
- [ ] Can coordinate multiple lasers
- [ ] Equip the kill laser and engage live targets
- [ ] spotted lanternfly
- [ ] mosquito
- Interface
- Data interface between camera and computer. Prefer the fastest interface available. Kare remarked that USB may be too slow.
- Night Vision
- Usually cameras with night vision include infrared LEDs and detect infrared wavelengths. At night these cameras produce black and white images.
- IR-cutoff
- Switches between day and night modes. Generally improves both day and night image quality but may cut back and forth at dawn and dusk.
- Distortion
- Fisheye lenses have high visual distortion. This will make it more difficult to map pixels to real world positions.
- Case / Mount
- A weather proof protective case would be a plus. Also we want a case that can be securely mounted to the platform. This matters because we’ll need to recalibrate every time the camera shifts.
- Resolution
- Prefer higher resolution, since we’re looking for bugs and bugs are small.
- Refresh Rate
- Prefer 60 fps or more, which is widely available.
- Focus
- Manual or autofocus.
- Single board computer components
- There are several cameras made for single board computers, such as Raspberry Pi and Jetson Nano. These connect to the board with a CSI port. Many provide night vision. Some have fisheye lenses. Most of these do not come with a case and may be difficult to mount. Most are manual focus.
- Security cams
- These are CCTV cameras. Many are USB; some use an analog cable. Most include night vision and IR LEDs with IR-cutoff. These usually come in a weather proof cases and would be easy to mount securely. Most are manual focus.
- Webcam
- These are usually USB. These come with a plastic case. Usually these can be mounted easily but aren’t weather proof. Some of these support night vision. Most are autofocus.
Eye-safe lasers can still damage eyes, but the damage is done to the cornea and will heal. Lasers that aren’t eye-safe cause damage to the retina which is permanent. Eye-safe lasers have wavelengths longer than 1400 nm.
IR Lasers are lasers with wavelengths in the infrared band (700 nm to 1 mm). These pose greater risk to eyes because since they aren’t visible, they don’t trigger the blink reflex, so eyes are more likely to be exposed for longer and sustain more damage.
Unfortunately all Eye-safe lasers are also IR. We should get laser safety goggles and use them whenever lasers are active.
Some links, will clean up later:
- https://en.wikipedia.org/wiki/Laser_safety
- https://www.rp-photonics.com/eye_safe_lasers.html
- https://seminex.com/lasers-and-eye-safety.aspx/
High powered lasers can pose a fire risk.
When we get the kill laser, we will need to test it with dry leaves in a controlled environment to determine how much power is required to cause ignition.
We have to prevent the system from sending beams into the sky or neighbors’ yards/houses. There should be software constraints but ideally also a physical shield should block the path of the laser.
Here is a list of laser examples by power. It is a good way of getting an idea of what different power ratings mean.
Using a laser more powerful than necessary increases risk of injury and fire. We need to determine the minimum power required for the laser diode that can be used to kill the target. According to Kare, “a few tens of millijoules, delivered within a few milliseconds, will cause most mosquitoes to expire within 24 hours.” This table shows how long we’ll have to keep beams of differing power on the insect to kill it.
Desired Energy (mJ) | Diode Power (mW) | Required Time (ms) |
---|---|---|
20 | 5 | 4000.000000 |
20 | 50 | 400.000000 |
20 | 100 | 200.000000 |
20 | 200 | 100.000000 |
20 | 5000 | 4.000000 |
30 | 5 | 6000.000000 |
30 | 50 | 600.000000 |
30 | 100 | 300.000000 |
30 | 200 | 150.000000 |
30 | 5000 | 6.000000 |
High power is required to get tens of mJ on target in single digit miliseconds, but the turret design may allow us to hit the target repeatedly in a fairly short time period. I think we should test with a 50 mW diode as an initial kill laser. 50 mW lasers are easy to find in the $15-$20 range.
- Hit the target with multiple beams that are a fraction of the desired kill power. The lasers will intersect at the target position, delivering full power.
- Hit the target with multiple short pulses. This allows us to move the beam between pulses, so repeated pulses from each laser will be unlikely to deliver much power to the same ground location and repeatedly hitting an eye would be highly unlikely.
- Disable the system if humidity is low to reduce risk of fire.
- Use a wide diameter beam to increase the chance of damaging a target’s delicate wings.
We should test with cheap low power (5 mW) lasers with wavelengths in the visible light band. These can be purchased for 10 for $10 in assorted colors, which will help if we are calibrating multiple turrets. These are equivalent to the lasers used in standard laser pointers.
This is the only way to be sure (https://youtu.be/nnHmUk_J6xQ). Note that this is assumed to be preferrable to nerve gas. However, the installation has a substantial dollar value attached to it and the action may not be authorized.
A candidate design could be a single laser turret. Optical image recognition might be used for target identification. A mounted laser would then be aimed and fired. A candidate use case would be deployment of the turret in a area with moderate to high density of spotted lanternfly. The turret would be manned and operated in a semi-autonomous mode. Semi-autonomous deployment would mitigate safety and weather challenges.
This is the solution described in detail in Kare, 2010.
Anonymous. (2020, July 4). “How Does Mosquito Laser Work? Here’s How To Build/Buy A DIY Killer.” PointerClicker. Retrieved July 4, 2020, from https://pointerclicker.com/how-mosquito-laser-defense-works/
Kare, J. (2010, Apr 30). “Backyard Star Wars.” IEEE Spectrum. Retrieved July 4, 2020, from https://spectrum.ieee.org/consumer-electronics/gadgets/backyard-star-wars
Myhrvold, N. (Feb 2010). “Could this laser zap malaria?” TED Talks. Retrieved July 4, 2020, from https://www.ted.com/talks/nathan_myhrvold_could_this_laser_zap_malaria
NIVDIA. (2020 Mar 19). “Getting Started with AI on Jetson Nano” NVIDIA Deep Learning Institute. Retrieved July 5, 2020, from https://courses.nvidia.com/courses/course-v1:DLI+C-RX-02+V1/info
Schiller, B. (2016, May 3). “What Happened To The Mosquito-Zapping Laser That Was Going To Stop Malaria?” Fast Company. Retrieved July 4, 2020, from https://www.fastcompany.com/3059127/what-happened-to-the-mosquito-zapping-laser-that-was-going-to-stop-malaria
Wikipedia contributors. (2020, May 16). Mosquito laser. In Wikipedia, The Free Encyclopedia. Retrieved 14:31, July 4, 2020, from https://en.wikipedia.org/w/index.php?title=Mosquito_laser&oldid=956918485
Wikipedia contributors. (2020, June 22). Spotted lanternfly. In Wikipedia, The Free Encyclopedia. Retrieved 14:43, July 4, 2020, from https://en.wikipedia.org/w/index.php?title=Spotted_lanternfly&oldid=963918748