The problems with cloud-only AI
Sending all sensor data to the cloud for processing is the most intuitive approach, but in a real product it usually runs into three issues:
- Latency: actions that need an instant response, such as reacting the moment a hand touches a surface, cannot wait for a round trip to the cloud.
- Bandwidth and cost: continuous, high-rate sensor data takes a lot of transmission and storage if everything is uploaded.
- Privacy: for data tied to the body, sleep or physiology, processing on the device first reduces how much data has to leave it.
What is EI (edge intelligence)
EI performs sensing and first-level decisions on the device or in the field, and sends results rather than raw data to the cloud only when needed. It does not replace cloud AI. The two share the work: the edge handles real-time, on-site decisions, while the cloud handles learning and integration that need large amounts of data.
Sensors are the foundation of edge intelligence
Compute resources at the edge are limited, and no algorithm can reach the right conclusion from poor data. So the characteristics of the sensor largely decide what EI can do: whether it conforms to curved surfaces, how thin it is, whether it measures many points at once, and how fast it responds.
JingSense's sensing solutions start from this point and run on two lines:
- Smart flexible pressure sensing: a thin, flexible pressure-sensitive film that measures pressure distribution as an array, turning "contact" and "force" into data an algorithm can use.
- Smart Optical Module: ToF time-of-flight and related optical technology that detects distance and gestures, sensing motion and approach without touching.
One senses "touch" and the other senses "approach". Each can be used on its own or together, depending on the product.
TAIPEI: how we see technology in the field
The TAIPEI on our homepage breaks down the four elements of putting technology to work:
| Letters | Stands for | Role in a solution |
|---|---|---|
| T | Technology | Choosing the right sensing principle and material, such as flexible pressure and optical sensing |
| AI | Artificial Intelligence | Using algorithms to turn sensor signals into meaningful results |
| P | Platform | Integrating sensing, firmware and applications into a complete solution that can be prototyped and mass-produced |
| EI | Edge Intelligence | Making decisions on the device and in the field, in the real use setting |
Four application directions to watch
These are application directions built on flexible pressure and optical sensing, shown to illustrate what EI looks like in a product. They are not specific customer cases.
- Electronic skin: giving machines, prosthetics or wearables a skin-like sense of pressure, including where and how hard something touches.
- Smart bedding and wearables: measuring body pressure distribution, and going on to sense physiological signals such as heart rate and respiration.
- Flexible heating: sensing and a thin-film heater on the same flexible material, adjusting heat to the user's contact and pressure.
- Touchless interaction: ToF and gesture sensing detect distance and motion, combined with projection so appliances, commercial displays and in-vehicle interfaces can be operated without touching.
From AI to EI
From AI to EI, JingSense brings intelligence into the field, empowering a new Taipei. If you are evaluating how to give your product a sense of touch, tell us about your application and requirements.
Interested in JingSense's sensing solutions, or want samples?
Tell us your application and we will recommend a suitable solution.