Panasonic Cycle Technology is using a microcontroller and edge AI development tool from Swiss firm ST Microelectronics for its TiMo A e-assisted bike.

The STM32F3 microcontroller (MCU) and STM32Cube edge AI development tool provide a tyre pressure monitoring system (TPMS) that leverages an AI function to improve rider safety and convenience.

Panasonic is a producer of e-assisted bikes with products for various uses in the Japanese market. Its TiMo A electric assist bicycle for school commuting runs an AI application on the STM32F3 to infer the tyre air pressures without using pressure sensors.

Based on information from the motor and the bicycle speed sensor, the system generates a warning to inflate the tyres if necessary.

The STM32Cube.AI enabled Panasonic to implement this edge AI function while fitting into STM32F3 embedded memory space. This function simplifies tyre air-pressure maintenance, which enhances rider safety and prolongs the life of tyres and other cycle components. It also helps reduce the cost and design work, as there is no need for additional hardware such as an air pressure sensor.

“We develop and manufacture e-assisted bikes with the mission of delivering environmentally friendly, safe and comfortable transportation, accessible to all,” said Hiroyuki Kamo, software development manager at Panasonic Cycle Technology. “The STM32F3 MCU provides cost competitiveness and optimal functions and performance for e-assisted bikes. By combining the STM32F3 MCU with STM32Cube.AI, we were able to implement the innovative AI function without the need to change hardware. We will continue to increase the range of models with AI functions and strive to fulfil our mission by leveraging STM’s edge AI.”

Marc Dupaquier, managing director for artificial intelligence at ST Microelectronics, added: “STM has been actively working on the global proliferation of edge AI in both hardware and software, providing edge AI to a wide range of products including industrial and consumer equipment. This collaboration marks a key step in our efforts, and we are delighted to have contributed to the first implementation of this AI function in Panasonic’s e-assisted bike. We will continue to propose AI use cases for diverse markets anywhere we can help to augment our life.”

The STM32F3 (www.st.com/en/microcontrollers-microprocessors/stm32f3-series.html) adopted for the TiMo A is based on the ArmCortex-M4 with a maximum operating frequency of 72MHz and features 128kbyte flash, along with various analogue and digital peripherals for motor control. In addition to the inflation warning function, the MCU determines the electric assistance level and controls the motor.

It leverages STM32Cube.AI to reduce the size of the neural network (NN) model and optimise memory allocation throughout the development of this AI function. STM32Cube.AI (www.st.com/content/st_com/en/campaigns/stm32cube-ai.html) converts NN models learned by general AI frameworks into code for the STM32 MCU and optimises these models. The tool optimised the NN model developed by Panasonic for the STM32F3 and implemented it in the flash memory, which has limited capacity.

STM (www.st.com) will showcase edge AI, including the STM32 MCU and a variety of AI development tools, at next month’s AI Expo in Tokyo (www.nextech-week.jp/hub/en-gb/lp/ai.html). The e-assisted bike and the motor unit from Panasonic (cycle.panasonic.com), which use the STM32F3 MCU and STM32Cube.AI, are also scheduled to be displayed at the expo.