Energy Efficiency in Embodied AI Systems

Résumé fourni par la source

The application of AI in physical objects, known as Embodied AI, is causing a big change in robotics, driverless vehicles, and smart gadgets. This chapter examines how to enhance the energy efficiency in AI, by adjusting systems and hardware and by making improvements to software. Improving AI algorithms is essential in the software part of making energy efficiency better. There are even algorithms that can change their energy use as needed by how difficult a job is or by the conditions around them. Split power designs, where parts of the system manage their power separately, support embodied AI well. It is possible to make data transfer even more efficient by using less power through communication methods. For embodied AI to be used more broadly, it must be designed to use less power. Energy-efficient embedded AI is important for practical uses and helps achieve global sustainability aim.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.