[Aldric Negrier] wanted to make 3D-scanning a person streamlined and simple. To that end, he created this voice-controlled 3D-scanning rig.
[Aldric] used a variety of hacking skills to make this project, and his thorough Instructable illustrates this nicely. Everything from CNC milling to Arduino programming to 3D-printing was incorporated into the making of this rig. Plywood was used to construct the base and the large toothed gear. A 12″ Lazy Susan bearing was attached to this gear to allow smooth rotation. In order to automate the rig, a 12V DC geared motor was attached to a smaller 3D-printed gear and positioned on the base. When the motor is on, the smaller gear’s teeth take the larger gear for a spin. He used a custom dual H-bridge motor driver made by a friend, which is connected to an Arduino Nano. The Nano is also connected to a Bluetooth module and an ultrasonic range finder. When an object within 1-35cm is detected on the rig for 3 seconds, the motor starts to spin, stopping when the object is no longer detected. A typical scan takes about 60 seconds.
This alone would have been a great project, but [Aldric] did not stop there. He wanted to be able to step on the rig and issue commands while being scanned. It makes sense if you want to scan yourself – get on the rig, assume the desired position, and then initiate the scan. He used the Windows speech recognition SDK to develop an application that issues commands via Bluetooth to Skanect, a 3D-scanning software. The commands are as simple as saying “Start Skanect.” You can also tell the motor to switch on or off and change its speed or direction without breaking form. [Aldric] used an Asus Xtion for a 3D-scanner, but a Kinect will also work. Afterwards, he smoothed his scans using MeshMixer, a program featured in previous hacks.
Check out the videos of the rig after the break. Voice commands are difficult to hear due to the background music in one of the videos, but if you listen carefully, you can hear them. You can also see more of [Aldric’s] projects here or on this YouTube channel.
Continue reading “Take A Spin On This Voice-Controlled 3D Scanning Rig” →
The Oculus Rift and all the other 3D video goggle solutions out there are great if you want to explore virtual worlds with stereoscopic vision, but until now we haven’t seen anyone exploring real life with digital stereoscopic viewers. [pabr] combined the Kinect-like sensor in an ASUS Xtion with a smartphone in a Google Cardboard-like setup for 3D views the human eye can’t naturally experience like a third-person view, a radar-like display, and seeing what the world would look like with your eyes 20 inches apart.
[pabr] is using an ASUS Xtion depth sensor connected to a Galaxy SIII via the USB OTG port. With a little bit of code, the output from the depth sensor can be pushed to the phone’s display. The hardware setup consists of a VR-Spective, a rather expensive bit of plastic, but with the right mechanical considerations, a piece of cardboard or some foam board and hot glue would do quite nicely.
[pabr] put together a video demo of his build, along with a few examples of what this project can do. It’s rather odd, and surprisingly not a superfluous way to see in 3D. You can check out that video below.
Continue reading “Seeing The World Through Depth Sensing Cameras” →
Charlotte’s chassis comes from as a kit, but the stock electronics are based on an Arduino – not something for a robot that needs to run computer vision apps. [Kevin] ended up using a Raspi for the controller and gave Charlotte eyes with an Asus XTION. Edit: or a PrimeSense sensor These sensors are structured light depth cameras just like the kinect, only about smaller, lighter, and have a better color output.
Hardware is only one half of the equation, so [Kevin] tossed the Arduino-based stock electronics and replaced them with a Raspberry Pi. This allowed him to hone his C++ skills and add one very cool peripheral – the
XTION depth camera.
To the surprise of many, we’re sure, [Kevin] is running OpenNI on his Raspberry Pi, allowing Charlotte to take readings from her depth camera and keep from colliding into any objects. The Raspberry Pi is overclocked, of course, and the CPU usage is hovering around 90%, but if you’re looking for a project that uses a depth sensor with a Pi, there you go.
Continue reading “Charlotte, The Hexapod With 3D Vision” →