A sophisticated Raspberry Pi 4 robot that observes terrain for danger could save human lives. Maker Aviv Butvinik tells Rosie Hattersley more in the latest issue of The MagPi, out now.
Imagine a robot that could traverse the unforgiving sands of the desert in the darkness of night, covering many miles at a time and stopping at given waypoints to surveil. It could collect video at these stop-points from a front-mounted controllable zoom camera equipped with infrared LEDs invisible to the naked eye, but illuminating for camera visibility. Then, it could beam this video data over an encrypted wirelessly connection to a distant location. Possible dangers surveilled, soldiers or explorers would be briefed on whether it was safe to proceed, potentially saving lives as well as time and money. This is the concept robot enthusiast and engineer Aviv Butvinik has been working on over the past two years. In December, he unveiled the second iteration of his Desert Eye surveillance robot with a YouTube video that shows off its ability to cope with harsh environments replete with sandstorms.
Desert Eye is also a great showcase for Raspberry Pi 4, which Aviv chose for its video processing capabilities – “a large consideration in this robot” – as well as controlling its motion and communications. Aside from the cameras, Raspberry Pi 4 controls the two main DC motors, a stepper motor so the camera can rotate while being stabilised, the GPS and wireless connectivity, and a three-axis sensor and fan.
Grand designs
Aviv had the sort of early school encouragement most of us can only dream of, getting to build model rockets, foam boats, and motorised balsa wood cars with his elementary school science teacher. “Those years instilled in me a love for science,” he says. Aviv’s father suggested he apply for the industrial design degree at the local university, where he eventually found his niche with a proposal for an engineering project that bagged him first place at the Design for Safety Symposium. This success led to Aviv taking up a post as a mechanical engineer and industrial designer creating automotive safety products. He says the project challenges, and solutions he’s exposed to make him think about “the physical forces, electrical engineering, and kinematics involved in designs.”
Aviv’s passion for robotics saw him design almost every aspect of Desert Eye himself, including the suspension dynamics, which he carried over from the first design to the current version. In an ideal world, he would have designed his own fasteners, motors, and even his own take on Raspberry Pi 4. Aviv sourced components from a range of online sources and used a 3D printing shop, but his best tip is to use VIAM open-source robot control software, which he favours for the online platform’s ease of use, and because Raspberry Pi can be flashed with a Lite version of the OS for the VIAM server, increasing its processing speed. “The VIAM team spent many hours helping me get Desert Eye’s programming just right and I hope to work with them again in the future on some of the more advanced visions I have for the robot.”
While Aviv was confident about Raspberry Pi 4 being powerful enough to process the video camera footage his surveillance robot captured, the infrared LEDs either side of the camera module had a tendency to wash out the footage. Moving them further away from the camera and angling them away from the ground so the light beams were narrower and a foot in front of the robot was a simple but effective fix.
Iterate and improve
Many of the challenges for Desert Eye’s design relate to its locomotion. The first version of the robot had an overly complex belt loop and dual sprocket system that drove the tank forward. “It looked and worked a lot like the suspension on a standard tank,” says Aviv, but the belts would sometimes fall off their sprockets due to road vibration. Aviv has replaced these with a lighter, 3D-printed flexible belt loop made from elastic that can tolerate much higher degrees of vibration and misalignment than chains.
Aviv also found that when the robot started up, there was “a huge drop in belt tension (extra slack), as the rockers all moved inwards.” He fixed this by adding a second rocker arm to the back rocker to compensate for the loss of tension.
The MagPi #127 out NOW!
You can grab the brand-new issue right now from Tesco, Sainsbury’s, Asda, WHSmith, and other newsagents, including the Raspberry Pi Store in Cambridge. You can also get it via our app on Android or iOS. And there’s a free PDF you can download too.
You can also subscribe to the print version of The MagPi. Not only do we deliver it globally, but people who sign up to the twelve-month print subscription get a FREE Raspberry Pi Zero Pico W!
