UBTECH's UGOT Is a Seven-in-One Kit for Building AI-Assisted Educational Robots

One kit for seven robot designs, from a two-wheeled balance-bot to a four-legged "spider" — and with local and remote AI powers, too.

Robotics specialist UBTECH has launched a crowdfunding campaign for the UGOT, a kit to build up to seven different robot designs — each of which offers integration with artificial intelligence platforms, including OpenAI's ChatGPT.

"UGOT is a versatile and innovative robotic kit that offers a wide range of capabilities and modes of operation," the company claims of its creation, aimed primarily at education. "With its powerful AI features, high computing power, and modular design, UGOT provides users with a simple and engaging way to experience advanced robotics technology."

UGOT aims to offer seven robots in a single kit, providing a modular design which can be tailored to particular tasks. (📹: UBTECH)

The UGOT kit offers the ability to build seven different designs, though not all parts are available in the lower reward tiers: a quadrupedal robot dog, a four-legged "spider" robot, a transforming car, a mechanum-wheeled omnidirectional car, an "engineer vehicle" which adds a gripper arm to the former, a two-wheeled self-balancing car, and a "wheeled and legged robot" which puts wheels on the bottom of bipedal legs.

Each robot includes a camera with first-person view beamed to a companion mobile app, visual tag recognition, human and motion tracking, and posture recognition. A 360-degree far-field microphone array links to a speech recognition system, promising integration with OpenAI's ChatGPT large language model (LLM), which gives the robot the ability to construct plausible responses to natural language interactions.

UBTECH showed off UGOT's capabilities at the World Robot Conference earlier this year. (📹: UBTECH)

Inside the robot's control system is a single-board computer system with a quad-core Arm Cortex-A55 processor running at up to 1.8GHz, an Arm Mali-G52 2EE graphics processor, and a neural network coprocessor offering a claimed 1 tera-operations per second (TOPS) performance at INT8 precision for local machine learning workloads including gesture, facial, text, and object recognition. There's 4GB ofd LPDDR4 RAM and an on-board nine-axis gyroscope, with two external sensor modules as standard: the camera and a time-of-flight (TOF) distance sensor.

On the software side of things UGOT provides a companion mobile app with first-person view and remote control plus a range of "AI"-powered games — including ones which rely on the on-board machine learning engine, such as gesture-recognition-based rock-paper-scissors. For those who would prefer to program the device themselves, the company provides a Scratch-based graphical programming environment and a Python software development kit (SDK).

The UGOT kit is now funding on Kickstarter, with early bird prices starting at $349 for the base kit good for building the self-balancing car and wheeled-and-legged robot or $519 for the full kit for all seven designs — both representing a claimed 42 per cent discount off the planned retail price.

All hardware is expected to be delivered in December this year, UBTECH says, with production already begun.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles