Lattice Blog


Infinite possibilities at the Edge – AI-enabled smart toys powered by low power FPGAs and AI solution stack

Infinite possibilities at the Edge – AI-enabled smart toys powered by low power FPGAs and AI solution stack
Posted 04/12/2022 by Hussein Osman

Posted in

The global smart toy market is projected to grow to nearly $70 billion by 2026, per Transparency Market Research. Smart toys take a variety of shapes, like electronic pets, robots, smart train kits, and more. When you hear the word toy, you likely automatically think these are for kids, however some are developed specifically targeting older generations. The current senior toy marketplace contains products that range in utility from providing companionship to increasing cognitive skills to facilitating communication.

A smart toy provides a technologically augmented environment for people to carry out tasks interactively, adapting to users’ actions. These toys often embed sensors with an image recognition chip that utilizes artificial intelligence (AI) to make this possible. To deliver a good overall user experience, these AI solutions need to be power efficient, easy-to-use, and have real-time responsiveness. The low power requirement is particularly critical as most of these toys are battery operated.

TOMY COMPANY, LTD., a Japanese entertainment company developing toys and merchandise, introduced its first AI-enabled smart toy called “Ami-Chan” in 2021. TOMY selected Lattice to provide a complete AI solution that combines our FPGAs and sensAI™ solution stack to accelerate and simplify the AI development and design process.

Image: TOMY, YouTube
Image: TOMY, YouTube

TOMY created Ami-Chan to act as a communication companion for seniors. The doll is designed with a vocabulary of approximately 1,600 Japanese words and utilizes facial recognition to identify each user and address them by name once it is registered in the toy’s memory bank. This is enabled by smart vision technology to recognize users and change conversations accordingly. The use of a Lattice FPGA and the sensAI stack provided TOMY with a compact, battery-powered Edge AI solution that could quickly and easily be incorporated into the design. Learn more about Ami-Chan in the company’s press release.

Lattice sensAI and TOMY video thumbnail
Watch Video: Lattice sensAI Powers Smart Toy by TOMY Company, LTD.

With the increasing number AI and machine learning (ML) technologies being incorporated into our everyday technology, there are seemingly infinite possibilities for new applications leveraging these functions for human recognition, face identification and key-phrase detection. Lattice’s AI and ML solutions are accelerating the development of this new genre of Edge products by providing AI processing functions that put key processing functionality closer to the device itself, rather than requiring data transmission to the cloud. In addition to the reduced latency and performance gained by doing so, this also helps protect users’ privacy.

Our comprehensive sensAI solution stack facilitates end-to-end AI/ML model training, validation and compilation. And, the recently added sensAI Studio design environment, a GUI-based tool, helps developers build accelerated ML applications more quickly. Configured with the tools in the Lattice sensAI solution stack, edge-computing designs can deliver real-time AI/ML performance while consuming very little power – as low as 1 mW to 1 W*.

sensAI Stack

With multiple low-power FPGA families and sensAI solution stack, we are working closely with our partners and customers to harness multimodal, intelligent sensor fusion, and AI/ML techniques to enhance user experiences and to significantly reduce operating power – from smart toys, smart city, smart factory, to PCs and client computing devices.

We encourage you to download our whitepaper to learn more about our tools and IP that help transform low-power FPGAs into intelligent AI/ML Edge computing engines. And, to learn more about Lattice sensAI and low power FPGAs, please contact us at or visit

* Based on Lattice iCE40 UltraPlus, CrossLink-NX, ECP5, and CertusPro-NX FPGAs