Business Introduction
SemsiML
In order to realize a convenient, safe and secure society, it is essential to have "sensors" that detect various situations and to utilize the data obtained from those sensors. We are introducing the "SensiML Analytics Toolkit", an embedded AI development support tool using AutoML (Automated Machine Learning) to make it easier to utilize sensor data.
- Embedded AI training is expensive and time-consuming, making it difficult to get started
- Considering introducing embedded AI, but lacking expertise outside of the field
- I want to add embedded AI functionality to the product's onboard microcontroller and perform status monitoring.
The SensiML analytics toolkit can solve these problems.
Benefits of adding embedded AI functions to system control microcontrollers
- Embedded AI functionality can be added without increasing hardware costs
- Compared to processing by a microcontroller alone, it is possible to distinguish between complex operating conditions.
Microcontroller only
easyMotion Detection
Simple stop/move status
Easy to detect
Basic microcontroller configuration image
Microcomputer + Embedded AI
detailMotion Detection
From time series data changes such as vibration
Complex condition determination possible
Microcontroller + embedded AI implementation configuration image
By using the SensiML Analytics Toolkit, anyone can easily create and optimize models, even if they do not have expertise in data science or machine learning. In addition, by automatically trying many algorithms and hyperparameters, you can efficiently find the optimal model, reducing time and costs and improving performance.
Traditional development methods
Development methodology with SensiML analytics toolkit
Since time series data from various sensors is available, it can be used in many applications.
Solution 1: Gesture Recognition
Advanced motion sensors such as IMUs, radars and passive IR grid array sensor ICs are being adopted into an ever-increasing number of IoT products. This creates exciting new opportunities to leverage these sensors for smart gesture-controlled interfaces. Devices that can benefit include:
- Consumer Wearables
- Smart Glasses
- Smart Home Appliances and Devices
- Wearables for Industrial Workers
- Toys and Gaming Accessories
Evaluation is easy with the SensiML analytics tool and QuickLogic evaluation board.
Made by QuickLogic
Evaluation Board
SensiML Analytics Toolkit (DataStudio)
Solution 2: Keyword Spotting
By integrating custom wake words and command phrases directly into products with built-in microphones, there is no dependency on third-party smart home hubs and voice control capabilities can be tailored to meet the user experience goals of your unique product.
SensiML's graphical UI pipeline allows easy and fast customization and rapid integration of command vocabulary into optimized deep learning ML embedded code available as a library or in C source format.
Solution 3: Anomaly Detection
Detects abnormalities in the robot.
The robot is equipped with an MPU board equipped with an acceleration sensor, and only normal acceleration sensor data is trained. If the robot comes into contact with an object or a person, unknown data is detected and the robot arm stops.
Things (people) come into contact
Unknown is detected and the robot arm stops.
For inquiries regarding this project, please contact:
Telephone enquiries
Contact by email
Inquiry FormMany in a variety of fields
Solution track record
We deliver email newsletters that provide useful information such as the latest news and industry trends. Subscription is free and easy, so please sign up!
If you have any questions or comments about Shinko Shoji, or would like to discuss our products or services, please feel free to contact us using the inquiry form below.