IngDan Kepler System – The Enabling Platform for AI Democratization
Figure 1. Illustration of Kepler System.
Kepler System (or simply K-System) is an open AI + IoT platform developed by IngDan Labs, which enables those developers with little AI/IoT/Hardware experience to quickly prototype their AI-based products, and helps them to solve many pressing problems that hinder wide adoption of AI technologies. The key features and functions K-system provides are:
1) Open Platform – Both hardware and software platforms of K-System are open. Developers can easily build their products based on these open platforms with little customization. It provides a hardware platform with built-in well-tuned voice module that can be readily connected to voice related AI services. The hardware board can be conveniently extended to connect with other IoT sensors, actuators, and devices through a service bus architecture. It provides a customized Android-based software platform tailored just for IoT devices and AI services, along with an easy to use software development platform and a set of unified APIs for connecting AI services. With cost-effective BOM, and scalability from simple (e.g. sensor unit) to complex (e.g. robotics) products, it can be regarded as the “Raspberry Pie” of the AI era, but with more functions and features way beyond just hardware and software platforms. It will significantly shorten the development cycle of AI related product for developers and innovators at different experience levels and help traditional industry to rapidly transform into AI-based industry.
2) Open Cloud Services – Unlike many other platforms, K-system does not bind any particular cloud services to its hardware and software platforms; and the developers never have to worry about which AI services they need to connect to at design time. Instead, it allows the developers to switch to different cloud services that best fit their needs even after design and even after market. This is achieved by developing a set standard and unified APIs for various AI service calls and building a thin cloud switch layer that switches service calls on demand to different cloud service providers the developers choose. For example, if a customer is not satisfied with a particular speech recognition service with Company X, he can easily switch to the similar service of Company Y through the AI service switch in the cloud, even after the K-system based product is shipped. In the future, this could also be the basis for standardization of APIs for AI services for all cloud service providers.