There are few things as frustrating when you’re trying to get some serious hacking done than intruders repeatedly showing up without permission. [All Parts Combined] has the solution for you, with a Kinect-based robotic sentry turret to keep them at bay.
The system consists of a Microsoft Kinect V2 connected to a PC, which runs an app to do all the processing, and outputs the targeting information to an Arduino over serial. The Arduino controls a simple 2-axis servo mount with an electric airsoft gun zip-tied to it. The trigger switch is replaced with a relay, also connected to the Arduino.
The Kinect V2 comes with SDKs that really simplify tracking human movement, and outputs the data in an easy-to-use format. [All Parts Combined] used the SDK in Unity, which allows him to choose which body parts to track. He added scripts that detect a few basic gestures, issues voice commands, and generates the serial commands for the Arduino. The servo angles are calculated with simple geometry, using XY coordinates of the target received from the SDK, and the known distance between the Kinect and turret. When an intruder enters the Kinect’s field of view it immediately starts aiming at the intruder’s heart, issues a “Hands Up!” command, and tells the intruder to leave. If the intruder doesn’t comply, it starts an audible countdown before firing. [All Parts Combined] also added a secret disarming gesture (double hand pistols), which turns the turret into an apologetic comrade. All it needs is a Portal-inspired enclosure.
It’s a fun project that illustrates how the Kinect can make complex computer vision tasks relatively simple. Unfortunately the V2 is no longer in production, having been replaced by the more expensive, developer focused Azure Kinect. We’ve covered several Kinect-based projects, including a 3D room scanner and a robotic basketball hoop.
[Adam]’s first robot arm build was a major disappointment, when the servos he had purchased for the build turned out to be terrible at holding an angle. With limited funds, he elected to improve on what he had, learning much about precision control techniques along the way. [Adam] taught himself how to implement industrial strength control loops using hobby hardware, by implementing additional encoders into servos and taking into account velocity and torque in addition to just position. With a magnetic encoder on the servo output shaft and a tiny optical encoder hand-built for inside the motor itself, much higher accuracy is achievable by allowing the control system to compensate for backlash.
If you’ve been following the Boston Dynamics project Spot, you’ve seen its capabilities and how we’re starting to see it being used in public more since its official release last year. But in a true display of how hobbyist electronics have been evolving and catching up with the big companies over the past few years, [Miguel Ayuso Parrilla] shows us his own take on the walking robot with CHOP, one of the finalists in this year’s Hackaday Prize.
Running the show are two main components, a Raspberry Pi 4B and an Arduino Mega. While the Mega interfaces with the servo controllers and provides filtering for sensors like the inertial measurement unit, the Pi takes all that data in and uses a series of Python scripts in order to determine the gait of the robot and which way the servos should move through an inverse kinematics model. To control the direction in which the body of the robot should accelerate, a Bluetooth remote controller sends commands to the Raspberry Pi.
We’re excited to see home-grown projects rise to this level of complexity, which would be mostly unheard of a few years ago in the maker scene, and only presented by large tech companies with tons of money to spend on research and development. There are other quadruped robots to inspire yourself on than Spot though, like this one with a spherical design and fold-out legs. Check this one in action after the break.
It’s not every day that we see someone trying something new with robot locomotion, but [kong]’s robot Rollyboi was made to do exactly that by mixing up the usual robot-wheel-motor layout. Instead of the robot using motors to drive wheels, Rollyboi is itself the wheel, and uses multiple simple arms (legs?) attached to hobby servo motors to propel itself. The idea is that the arms swivel out one at a time to roll the robot along as needed.
It’s a novel idea, but how well does it work in practice? The first version was blind and mechanically unstable, with no idea which way was up and therefore no way to effectively control which arm needed to be extended, but was nevertheless able to roll along. The next version implemented a simple control system: buttons installed along the outside rim let the robot know how it is moving and which arm to extend next. With two sets of arms (one on each side) the robot becomes capable of executing simple turns by extending one arm more than the other.
In the end, Rollyboi could move but still lacks a means to perceive and navigate its environment. This is made more challenging by the fact that the robot’s body (and therefore any sensors mounted to it) would be in constant motion as the robot moves. Still, it’s interesting to see how far the idea went using only simple hardware, and its motion gives off a certain radial solenoid engine vibe. You can watch a brief video below.
When the Skynet baseball bot swarms attack, we’ll be throwing [Carl Bugeja] some dirty looks for getting them started. He’s been working on 4B, a little quadruped robot that can transform itself into a sphere almost perfectly.
Before [Carl] was distracted by the wonders of PCB actuators more than a year ago, he started working on this little guy. He finally found some time to get it moving on its own, and the preliminary results look promising to say the least. Inside the 6 cm sphere is a total of 12 servos, 3 for each leg. All of the mechanical parts were 3D printed in nylon on an SLS machine, and the custom PCB has a BLE microcontroller module, an IMU and IR proximity sensors onboard. Everything is open source with all the files available on the Hackaday.io project page.
The microcontroller runs a full inverse kinematic model, so only the desired tip and base coordinate for each leg is input and the servo angles are automatically calculated. Ultimately [Carl] aims to have the robot both walking and rolling controllably. So far he’s achieved some degree of success in both, but it still needs some work (see the videos below. We’re eager to see what the future holds for this delightfully creepy bot.
[Nick Bild]’s latest hack helps you find objects (or people) by locating their position and tracking them with a laser. The device, dubbed Artemis, latches onto your eyeglasses and can be configured to locate a specific object.
Images collected from the device are streamed to an NVIDIA Jetson AGX Xavier board, which uses a SSD300 (Single Shot MultiBox Detection) model to locate objects. The model was pre-trained with the COCO dataset to recognize and localize 80 different object types given input from images thresholded in OpenCV. Once the desired object is identified and located, a laser diode activates.
Probably due to the current thresholds, the demo runs mostly work on objects placed further apart against a neutral background. It’s an interesting look at applications combining computer vision with physical devices to augment experiences, rather than simply processing and analyzing data.
The device uses two servos for controlling the laser: one for X-axis control and the other for Y-axis control. The controls are executed from an Adafruit Itsy Bitsy M4 Express microcontroller.
Perhaps with a bit more training, we might not have so much trouble with “Where’s Waldo” puzzles anymore.
[Will] wanted to build some animatronic eyes that didn’t require high-precision 3D printing. He wound up with a forgiving design that uses an Arduino and six servo motors. You can see the video of the eyes moving around in the video below.
The bill of materials is pretty simple and features an Arduino, a driver board, and a joystick. The 3D printing parts are easy to print with no supports, and will work with PLA. Other than opening up holes there wasn’t much post-processing required, though he did sand the actual eyeballs which sounds painful.