Automotive
How a sound solution is making our roads safer
Reading a short text message on the road takes about five seconds. At 80 km/h, this is equivalent to driving the length of a football field blindfolded. The potential consequences can be fatal for the car occupants or other road users. Thanks to Autotrak's innovative AI-powered camera, truck drivers and employees driving for work can be helped to kick this perilous habit. The fleet management specialist has developed an advanced, yet cost-effective, real-time audio warning system that discreetly nudges drivers to put down their phones, with the goal of preventing accidents and significantly enhancing road safety.
Challenge
- Create a reliable but cost-effective audio warning system able to detect when the driver is using a phone
- Ensure efficient development time mitigating the learning curve associated with adoption of a new processing platform
Solution
- Leverage the STM32N6, ST’s high-performance MCU with AI acceleration, to benefit from extended computing capabilities, and achieve MPU-like results at reasonable costs.
- Quickly translate complex AI models into usable code using ST’s edge AI suite
Impact
- A system that can be broadly deployed to reduce one of the main causes of distraction
- Faster design & development time
Computer vision and AI at the edge for real-time decisions
On the road, every second counts. For years, designers aiming to enhance camera vision with real-time decision-making faced a daunting dilemma. They could either opt for an expensive processing solution, which drove up the final solution's cost, or settle for affordable cloud-based computing, which made real-time decisions a distant dream. In 2024, the landscape shifted when ST unveiled its STM32N6 MCU, the first microcontroller with AI acceleration, delivering microprocessor-like performance in compact embedded systems. For Autotrak, the opportunity was too compelling to ignore. The fleet management specialist emerged as one of the early testers of the STM32N6 MCU.
“
ST has been a trusted partner for more than a decade, supplying a range of chips for our tracking and monitoring equipment. The STM32N6 MCU takes our relationship to a different level.
”
Gavin Leask, Engineering Director, Autotrak
Who ever thought a microcontroller could do all that?
With the STM32N6’s embedded image signal processing, Autotrak’s engineering team was able to extract the best image from their medium-resolution camera without using a more expensive processing solution like a microprocessor. The STM32N6’s sophisticated AI image classification allowed their standard camera to analyze and categorize millions of driver gesture images stored in the device’s expansive memory. All this directly in the device and in the blink of an eye. And the success story doesn’t stop here. Unlike traditional microprocessors, this high-performance MCU is resilient to high temperatures. This robustness enabled Autotrak to mount their advanced driver-assistance system (ADAS) on the vehicle’s windscreen, ensuring reliable performance under the most demanding conditions.
“
The STM32N6 MCU has enabled us to develop a cost-effective, compact, low-power ADAS solution that operates at high temperatures, unlike traditional MPU solutions.
”
Gavin Leask, Engineering Director, Autotrak
How ST makes AI in computer vision more accessible
Thanks to the STM32N6's built-in neural processing unit, also called Neural-ART Accelerator, developers are now able to unleash their creativity without overly worrying about the costs. But the best part for Autotrak has been how ST supported their designers every step of the way. From ready-to-use model optimizations and code conversion to a powerful network compiler and deployment tools, the team was able to work smarter, not harder. And save precious time. Autotrak launched the prototype in under 3 months, continuing their long legacy of 'More than just tracking,' by prioritizing customer safety on the road.
“
The ST Edge AI Suite with wide ranging tools, including its built-in optimizing AI network compiler, alleviates the need for a complex and lengthy design cycle.
”
Gavin Leask, Engineering Director, Autotrak
