Sybil Is a Solar-Powered, TensorFlow Lite-Equipped Smart Gardening Robot
Building on an AWS DeepRacer prototype, Sybil includes stereo cameras and a LiDAR sensor with mapping, planting, and weeding support.
Sterling Robotics is looking to fund production of a TensorFlow Lite-enabled gardening robot, driven by solar power: Sybil.
"Sybil is the smartest garden companion around," claims Sterling Robotics' Venessa Mann. "Sybil can plant seeds, and keep weeds away from your garden, all on her own. Completely chemical free and no Wi-Fi needed. Over time, Sybil learns what spots of the garden need more attention than others, further maximizing battery usage."
The smarts extend beyond learning where in the garden needs the most work, too. A pair of front-mounted cameras are linked to a TensorFlow Lite neural network designed to offer autonomous navigation and to identify weeds — preventing it from accidentally chopping down the plants you're trying to grow. Additional navigation and obstacle-avoidance capabilities come courtesy of a LiDAR sensor.
"The original prototype was build using the base of an AWS [Amazon Web Services] DeepRacer," Mann explains. "In this version we replaced the compute unit with our own." Formal specifications of the on-board computer, however, have not been released.
As well as mapping a garden, Sybil is claimed to identify where water is required, automatically plant seeds loaded into its on-board hopper, and chip down weeds with an integrated cutting tool — before marking where it was growing on its map, to keep monitoring for any regrowth.
The company has launched a Kickstarter campaign, where physical rewards are available starting at $297 for an early-bird Sybil with five replacement cutters and four pins for setting up the corner boundaries of its patrol area. Shipping is expected to take place by the end of the year.