AI Based Hand-held Device

The Scenario

With the advancement of AI+IoT and communication technology, we can see more and more smart handheld devices appearing in our daily life. Such smart handheld devices are the application transformation of smartphones from personal use to industrial use. It has several Main features: 1. Equipped with the APP required for industrial applications, which is the main factor for the industrialization of handheld devices 2. Data collection capability and diverse network services: Collect on-site data by Camera/NFC/RFID and other technologies, through BT/ Wifi is used as a local network connection, and handshakes data with the cloud (backstage) through 4G/5G 3. A stronger mechanism design is required to cope with a variety of use environments, such as more waterproof, shock-resistant, drop-resistant, etc. This way The device has been seen everywhere in our life, such as the device used by the police to search for license plates, the device of the suspect, the device used by the waiter in the restaurant to order food, the scanning device used by the warehouse to obtain the information of the goods, and so on.

In recent years, due to the improvement of AI technology, the application of Edge AI has also appeared in a large number of smart handheld devices, such as license plate recognition (parking fees), face recognition (transaction security), object recognition (warehousing applications), etc.

Aidc device

The Complication

Keeping a stable networking capability is a very important requirement for the use of smart handheld devices. The main functions of such devices are completed through the connection with the background. Therefore, how to maintain the connection with the background is constantly being studied. In addition, with the development trend of multi-tasking services of such devices, there are higher requirements for the basic core computing power, the processing speed of data per unit time, and even the speed of data transmission in the network.

The introduction of edge AI applications also makes such devices have a demand for AI computing power. However, the development and integration of applications has always been the biggest bottleneck in the realization of AI applications. How to quickly deploy existing algorithms on devices can make them faster and faster. One step to bring products with AI services to the market is also the current challenge for such devices.

However, in handheld devices, like your mobile phone, power saving is a very important requirement, but, as mentioned earlier, basic computing/AI acceleration is also very important, so how to strike a balance between performance and power consumption is another important topic for such devices.

Finally, this type of device is hand-held after all, and there are restrictions on size/weight. Therefore, it is hoped that the core is an integrated solution that provides high-efficiency computing cores, and has AI acceleration and network functions. More importantly, the price is Smart handheld devices are acceptable. There is currently a lack of such solutions on the market.

MediaTek Solution

In response to the needs of smart handheld devices, MTK MT8788 will be the perfect solution

MT8788 provides 8-core CPU (4 x ARM A73 2.0 GHz + 4 x ARM A53 2.0 GHz) data processing capability, and has sufficient performance for various smart handheld device applications, and built-in independent APU (AI Acceleration Processor), It has 0.7 TOPS AI computing power, and uses the standard and commonly used tensorflowlite as the integration interface, which greatly reduces the development resources required for integration, and is easy to implement various AI services and applications.

In terms of network connection capability, it can provide 4G Cat7 and support 5G+5G dual card dual standby, greatly improving the stability and coverage of the connection

With such a highly integrated, multi-tasking, and efficient solution, its power consumption is less than 7W, which allows smart handheld devices to have better power usage efficiency and longer working hours.

The Result

After introducing the MT8788, a leading brand of smart handheld devices in the United States (retail field) obtained the following benefits. Because of the convenience of MT8788 AI integration, the development time was shortened by about 15%, and the time to market for products was accelerated, leading the industry. In actual use, due to the low power consumption of MT8788, the use time is extended by 20%, and the introduction of 5G+5G solution reduces the unstable connection between the product and the background by 17%. Overall, the use of the whole machine On, increase the work efficiency by 20%.

How to get started

Visit the Product page and other materials related to this case study.