Will Cogley's Coglet Is a 3D-Printable, Expressive, Twin-Camera Companion Bot
Compact companion robot includes nine degrees of freedom and a dual-microcontroller layout, with animations programmed in Python.
Maker Will Cogley is crowdfunding for an open source, 3D-printable companion robot programmable in Python and with online and offline operating modes: the Coglet.
"We believe robots should be expressive and full of life, not just minimalist shells and monochrome screens," Cogley says of the inspiration behind his semi-eponymous companion bot. "Coglet is a robotics kit for those of us who grew up with WALL-E, R2-D2 and Baymax — robots that communicated through movement, mechanics, and personality. Coglet combines intricate mechanical design, expressive animation, and modern electronics into a single maker kit. Unlike other screen-based assistants, Coglet moves, reacts, and behaves with life and personality."
The platform underpinning the expressive Coglet offers a total of nine degrees of freedom, though is designed as stationary desk-bound companion rather than an ambulatory roving robot: its eyes, ears, mouth, and eyelids all move under programmatic control, along with its head via neck roll and pitch and base yaw. Designed, Cogley says, as a mechanical project as much as an electronics one, this is all handled by a series of 16 gears and a total of 60 3D-printed parts.
Inside the robot's shell is a custom control board dubbed the CogNog, with two microcontrollers: a Raspberry Pi RP2040 driving the device's nine servos plus an optional 10th and handling animation tasks, and an Espressif ESP32-S3 communications coprocessor delivering Wi-Fi and Bluetooth Low Energy (BLE) connectivity along with audio input and output capabilities. Each eye, meanwhile, has its own smart camera embedded: one is a Seeed Studio Grove Vision AI Module V2 running pre-trained face, person, and object detection models on-device with the option of supporting custom models locally, while the other is linked to the ESP32 for streaming video or outputting to a cloud service via the open source XiaoZhi AI platform for more power-hungry analysis.
Cogley is funding the project on Kickstarter, with physical rewards starting at £25 (around $33) for a CogNot board, speaker, and microphone, £125 (around $167) for all but the 3D-printed parts, or £250 (around $334) for a full kit; all hardware is expected to ship between August and September this year, though as with all crowdfunding campaigns that's not a guarantee. "Although all of our planned features are working," Cogley says of the plan for fulfillment, "we plan to speed things up and make a few more mechanical tweaks so that Coglet can be quieter, more robust and easier to assemble."
Files for the current prototype, meanwhile, are available on GitHub under an unspecified license.
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.