Conclusion

Moore’s Law suggests that processor speeds double every 18 months to two years. According to a study done by Stanford University and published in the AI Index 2019 annual report, the speed of artificial intelligence (AI) is outpacing Moore’s Law. The study found that post-2012, computing power has been doubling every three to four months.

Because of the challenges in defining proper hardware acceleration, many companies have taken the route of enabling AI with tools and software on general purpose controllers and monitoring the evolution.While embedded AI can make smart data from big data, creating a system for efficient AI workload processing should begin with understanding the current and future evolution of AI applications and the specific software and hardware changes required to support those computations.

Designing lightweight, real-time, and energy-efficient architectures for embedded AI on the edge is an important research direction to pursue and is likely to continue to gain further momentum in the coming years. As artificial intelligence and its applications become more widespread, the race to develop cheaper and faster chips supporting embedded AI is likely to accelerate as well. It will be the AI hardware-software co-design and the associated tool chains that will provide the foundation for the next big developments in embedded AI.