The build relies on a Raspberry Pi kitted out with its usual camera for machine vision purposes. It uses the Viam robot toolkit, which runs a machine learning model to detect pets and humans on the camera feed. The guardian then tracks any pets or humans that show up by turning its head, and thus the camera, with a servo controlled by a PWM signal via the Raspberry Pi’s GPIO pins. It’s all wrapped up in a nicely-decorated 3D printed model that really does look like something straight out of Breath of the Wild.
Sentry projects are a great way to learn about electronics, mechanics, and image processing techniques. It’s funny to see how advanced and complicated these projects were fifteen years ago, compared to how easy they are today with modern machine learning libraries. How times change!
The build is one that leverages typical 3D printer components to get the job done. A Minitronics 2.0 board is used to run the show, packing a 40 MHz SAMD21 microcontroller for plenty of grunt. It’s Arduino compatible too, making it easy to program. It’s combined with NEMA17 and NEMA23 steppers and an external driver board to slew the gun towards a target. Target detection is via a RPLIDAR A1, which detects the range of nearby objects. This data is used to calculate the pan angle and tilt required to hit the target with a stream of water, fired by a relay-controlled solenoid.
What has dual compressed-air cannons, 500 roll-on deodorant balls, and a machine-learning brain with a bad attitude? We didn’t know either, until [Leo Fernekes] dropped this video on his autonomous robot sentry gun and saw it in action for ourselves.
Now, we’ve seen tons of sentry guns on these pages before, shooting everything from water to various forms of Nerf. And plenty of those builds have used some form of machine vision to aim the gun onto the target. So while it might appear that [Leo]’s plowing old ground here, this build is chock full of interesting tips and tricks.
It started when [Leo] saw a video on TensorFlow basics from our friend [Edje Electronics], which gave him the boost needed to jump into an AI project. The controller he ended up with looks for humans in the scene and slews the turret onto target, where the air cannons can do their thing. The hefty ammo is propelled by compressed air, which is dumped into the chamber using a solenoid valve with an interesting driver that maximizes the speed at which it opens. Style points go to the bacteriophage T4-inspired design, and to the sequence starting at 1:34 which reminded us of the factory scene from RoboCop.
[Leo] really put a ton of work into this project, and the results show. He is hoping to get an art gallery or museum to show it as an interactive piece to comment on one possible robot-human future, presumably after getting guests to sign a release. Whatever happens to it, the robot looks great and [Leo] learned a lot from it, as did we.
What is this world coming to when you can’t even enjoy sitting in your living room without some jamoke flying a drone in through the window? Is nothing sacred? Won’t someone think of the children?
Apparently [Drew Pilcher] did, and the result is this anti-drone sentry gun. It’s a sturdily built machine – one might even say it’s overbuilt. The gimbal is made from machined steel pieces, and the swivels are a pair of Sherline stepper-controlled rotary tables with 1/40 of a degree accuracy selling for $400 each. Riding atop that is a Nerf rifle, which is cocked by a stepper-actuated linear slide, as well as a Kinect for object tracking. The tracking app is a little rough – just OpenCV hacked onto the Kinect SDK – but good enough for testing. The gun tracks as smoothly as one would expect given the expensive hardware, and the auto-cocking feature works well if a bit slowly. Based as it is on Nerf technology, this sentry is only indicated for the control of the micro-drones seen in the snuff video below, but really, anyone afflicted by indoor infestations of Phantoms or Mavics has bigger problems to worry about.
Over-engineered? Perhaps, but it’s better than letting the menace of indoor drones go unanswered. And it’s far from the first sentry gun we’ve seen, targeting everything from cats to squirrels using lasers, paintballs, and even plain water.
Thieves beware. If you prowl around [Matthew Gaber]’s place, you get soaked by his motion activated super-squirter. Even if he’s not at home, he can aim and fire it remotely using an iPhone app. And for the record, a camera saves photos of your wetted-self to an SD card.
The whole security system is handled by three subsystems for target acquisition, photo documentation, and communications. The first subsystem is centered around an ESPino which utilizes a PIR sensor to detect motion. It then turns on a windscreen washer pump and uses pan and tilt servos to squirt water in a pattern toward the victim.
The target acquisition hardware also sends a message to the second subsystem, an ArduCAM ESP8266 UNO board. It takes a burst of photos using an ArduCAM Mini Camera mounted beside the squirter outlet. The UNO can also serve up a webpage with a collection of the photos.
The final subsystem is an iPhone app which talks to both the ESPino and the UNO board. It can remotely control the squirter and provide a video feed of what the camera sees.
One detail of the build we really enjoyed is the vacuum relief valve he fabricated himself. It prevents siphoning through the pump when it’s not on. Don’t miss a demo of the squirter in action after the break.
We love taking on new and awesome builds, but finding that second part (the “awesome”) of each project is usually the challenge. Looks like [Nathan Seidle] is making awesome the focus of the R&D push he’s driving at Sparkfun. They just put up this safe cracking project which includes a little gamification.
The origin story of the safe itself is excellent. [Nate’s] wife picked it up on Craig’s List cheap since the previous owner had forgotten the combination. We’ve seen enough reddit/imgur threads to not care at all what’s inside of it, but we’re all about cracking the code.
The SparkX (the new rapid prototyping endeavor at Sparkfun) approach was to design an Arduino safe cracking shield. It has a motor driver for spinning the dial and can drive a servo that pulls the lever to open the door. There is a piezo buzzer to indicate success, and the board as a display header labeled but not in use, presumably to show the combination currently under test. We say “presumably” because they’re not publishing all the details until after it’s cracked, a process that will be live streamed starting Wednesday. This will keep us guessing on the use of that INA169 current sensor that plugs into the safecracking shield. There is what appears to be a reflectance sensor above the dial to keep precise track of the spinning dial.
Electrically this is what we’d expect, but mechanically we’re in love with the build. The dial and lever both have 3D printed adapters to interface with the rest of the system. The overall framework is built out of aluminum channel which is affixed to the safe with rare earth magnets — a very slick application of this gear.
The gamification of the project has to do with a pair of $100 giveaways they’re doing for the closest guess on how long it’ll take to crack (we hope it’s a fairly fast cracker) and what the actual combination may be. For now, we want to hear from you on two things. First, what is the role of that current sensor in the circuit? Second, is there a good trick for optimizing a brute force approach like this? We’ve seen mechanical peculiarities of Master locks exploited for fast cracking. But for this, we’re more interested in hearing any mathematical tricks to test likely combinations first. Sound off in the comments below
[Brittliv] made the mistake of getting her friends into Nerf weaponry, and so began the race to mutually secured destruction via foam darts. She may have the upper hand in this war, because her Nerf Vulcan sentry gun is both incredibly powerful and is able to be operated autonomously with a webcam featuring a friend or foe identification system.
The azimuth and elevation mount for the gun is made out of plywood, with each axis controlled by a single servo attached to an Arduino. Of course a stock Nerf gun would be fairly boring, so [BrittLiv] increased both the voltage going to the gun’s motor and the strength of the gun by replacing a 2kg spring with a 5kg spring.
Targets are tracked with a webcam using Processing and a bit of code from Project Sentry Gun. One interesting feature is a friend or foe tracking system; if the gun sees someone wearing a t-shirt with the Instructables logo, the target is identified as a friend and is not brutally mowed down with plastic darts.