The automotive industry is moving rapidly forward with automation and electrification, making vehicles part of the growing IoT landscape. Artificial intelligence (AI) and machine learning are helping carmakers transform the huge amounts of data collected by vehicles into actions that improve the safety and comfort of the passengers, along with the performance and efficiency of the vehicle. In this article, we look at how Edge AI can provide further improvements.
Maintenance is a crucial part of car ownership. In the past, maintenance, like oil and filter changes, was conducted based on time or distance traveled, for example, a service every 10,000 kilometers or every six months. This is known as planned maintenance, and while it is sufficient in most instances, unexpected issues would go unnoticed if they occurred well before the scheduled service date.
New cars now contain a multitude of sensors, instruments, cameras, and even microphones that capture data associated with the car and its parts. This complex data can then be leveraged along with past service history in combination with AI insights to provide a vehicle owner with live or even pre-emptive notifications of faulty components or errors. Not only can this system suggest which parts need maintenance or replacement, but it can also automatically instill methods to reduce the possibility of catastrophic or permanent damage. This is called predictive maintenance.
The implementation of AI-based predictive maintenance functionalities hinges on the processing of many data points, which has been typically carried out within remote data centers, i.e., cloud computing infrastructures. In recent years, a shift in AI data processing is occurring from remote data centers to local infrastructures, commonly known as Edge AI infrastructures.
Nowadays, machine learning functions can be executed within sensors and hardware devices, such as the car’s OBU (on-board unit). This accelerates fault detection and the extraction of maintenance insights, helping drivers and automotive manufacturers make more timely decisions about the vehicle’s service and maintenance. Specifically, predicting and anticipating failures in real-time is possible, followed by suggested actions that lead to their avoidance.
In general, Edge AI enables real-time maintenance functionalities while contributing to more power-efficient vehicle operations. This is because local (edge) processing limits the amount of data transferred to remote locations, resulting in a significantly reduced CO2 footprint. This approach better aligns with the environmental performance goals of electromobility.
Another example of an Edge AI application is security. After vehicle thefts steadily decreased over decades, 2020 saw a rise, undoubtedly due to various factors, including the pandemic and the state of the economy. (10)
Cameras have been shown to be an effective automotive theft deterrent. (11) However, monitoring an empty car’s surroundings, 24/7 has its challenges. For example, what constitutes a threat, and how can the break-in be prevented? AI technology comes to the rescue with image processing and motion sensor capabilities, such that the system can assess and respond to risks immediately, e.g., by flashing interior lights as a warning. Some might think this feature in a vehicle would drain the battery. While that’s typical of batteries in cars with internal combustion engines, advances in electric vehicle technology have yielded ultra-low power and high-performance solutions.
Thanks to Edge AI infrastructures, predictive maintenance, and security functionalities can be added to the vehicle with a minimal power penalty, while previously adopted Advanced Driver Assist Systems (ADAS) can be improved. For example, by having AI capabilities on edge devices, like the OBU, modern vehicles can instantly identify and perceive dangerous conditions, such as obstacles, and react with very low latency. The driver can then navigate safely, or autonomous safety actions, such as breaking, can engage.
The implementation of real-time AI functions is instrumental in self-driving cars. They must evaluate the driving context and make decisions many times per second. In this direction, Edge AI functions are needed to analyze information within milliseconds very close to the field without the latency of cloud transmission.
Note also that a vehicle’s security and safety functionalities can be reinforced when understanding and evaluating the surrounding condition, as well as the state of health or alertness of the driver. In such cases, potentially sensitive data needs to be guarded. Edge AI helps to provide an extra level of data protection during processing, enabling outline outcomes to be transmitted outside of the vehicle, while keeping private data, such as images and audio, secure.
Syntiant has developed a family of cutting-edge deep learning processors to design for AI at the physical edge. The processors facilitate the execution of AI functions inside vehicles and obviate the need for sharing data outside the automobile. This is foundational for low-latency, power-efficient, and highly secure AI. Headquartered in California, US, the company was founded in 2017 with the mission to make Edge AI pervasive by providing processors with high-performance-to-power ratios and supplementing them and deep learning algorithms to enable a seamless interface between the analog and digital worlds. Syntiant’s processors address various automotive needs efficiently, like predictive maintenance and vehicle security. The powerful always-on AI can easily run on a 12V car battery for months without a perceptible drop in battery charge. Furthermore, multiple applications can run simultaneously at significantly lower power, such as people monitoring, intrusion detection, and voice-controlled infotainment systems.
Edge AI technology by Syntiant brings reliable security, comfort, and connection to the driver at a low power cost. In this way, it also empowers automotive manufacturers to develop and deploy advanced AI functionalities that will shape the future of transport, including some of the long-anticipated autonomous driving vehicles.
Syntiant combines advanced silicon solutions and deep learning models to provide ultra-low-power, high performance, deep neural network processing for edge AI applications across a wide range of consumer and industrial use cases, from earbuds to automobiles. The company’s Neural Decision Processors (NDP) are optimally designed to deploy deep (machine) learning models on the edge, where power and area are often constrained. Syntiant’s NDP solutions can equip, every device, from earbuds and doorbells to automobiles and healthcare wearables, with powerful deep learning capabilities, enabling real-time data processing and decision making, with near zero latency, enabling secure and private artificial intelligent solutions.
Lawton G. Edge AI: The Future of Artificial Intelligence and Edge Computing. [Online]. 2021 [cited 2022 September 15]. Available from: https://www.techtarget.com/searchenterpriseai/definition/edge-AI.
The Linux Foundation. State of the Edge. [Online]. 2021 [cited 2022 September 15]. Available from: https://www.lfedge.org/wp-content/uploads/2021/08/StateoftheEdgeReport_2021_r3.11.pdf.
QuantumBlack. The state of AI in 2020. [Online]. [cited 2022 September 15] Available from: https://www.mckinsey.com/business-functions/quantumblack/our-insights/global-survey-the-state-of-ai-in-2020.
Vyas K. IT Business Edge. [Online]. 2021 [cited 2022 September 15] Available from: https://www.itbusinessedge.com/data-center/developments-edge-ai/.
Gartner. Gartner Identifies Top Five Trends in Privacy Through 2024. [Online]. 2022 [cited 2022 September 15] Available from: https://www.gartner.com/en/newsroom/press-releases/2022-05-31-gartner-identifies-top-five-trends-in-privacy-through-2024.
Yeruva V. Edge Computing’s Applications In Autonomous Driving And Business At Large. [Online]. 2021 [cited 2022 September 15] Available from: https://www.forbes.com/sites/forbestechcouncil/2021/09/27/edge-computings-applications-in-autonomous-driving-and-business-at-large/?sh=578dbc9e96aa.
Arena F, Collotta M, Luca L, Ruggieri M, Termine FG. Predictive Maintenance in the Automotive Sector: A Literature Review. Math. Comput. Appl. 2022; 27(2).
Rafie M. Edge AI Computing Advancements Driving Autonomous Vehicle Potential. [Online]. [cited 2022 September 15] Available from: https://www.gsaglobal.org/forums/edge-ai-computing-advancements-driving-autonomous-vehicle-potential/.
Gonfalonieri A. How to Implement Machine Learning For Predictive Maintenance. [Online]. 2019 [cited 2022 September 15] Available from: https://towardsdatascience.com/how-to-implement-machine-learning-for-predictive-maintenance-4633cdbe4860.
Insurance Information Institute. Facts + Statistics: Auto theft. [Online]. [cited 2022 September 15] Available from: https://www.iii.org/fact-statistic/facts-statistics-auto-theft.
The New York Times. The Crime Rate Drops, and a City Credits its Embrace of Surveillance Technology. [Online]. 2007 [cited 2022 September 15] Available from: https://www.nytimes.com/2007/05/29/nyregion/29east.html.