Lecture
Industrial AI for the Edge
- at -
- Hall C6C6.501
- Language: English
- Type: Lecture
Lecture description
Edge AI is rapidly becoming a foundational technology in industrial systems, where real‑time decision‑making, reliability, and long product lifecycles are critical. Unlike cloud‑based AI, industrial Edge AI must operate within strict constraints on power, memory, latency, and safety while integrating into complex, long‑lived platforms. These constraints fundamentally shape how AI can—and should—be designed for industrial use. As this talk will explore, successful industrial Edge AI depends less on model accuracy alone, and more on system‑level design choices. Challenges such as hardware heterogeneity, deterministic behavior, data availability, and maintainability often determine whether Edge AI delivers real value. This session is motivated by the need to move beyond generic AI narratives and instead focus on practical design trade‑offs, co‑optimization of hardware and algorithms, and engineering realities that govern deployment at the industrial edge. To that end, we will look more specifically on the topic of running Edge AI in production cells on assembly lines to highlight how it can be used to reduce production faults, improve yield and maintain high levels of quality control. This talk is aimed at embedded engineers, machine learning engineers, system architects, and technical leaders working in manufacturing. It is particularly relevant for teams evaluating or deploying Edge AI in production systems and seeking a realistic understanding of design constraints and architectural decisions. Attendees will gain a clearer framework for designing Edge AI solutions that are robust, scalable, and aligned with industrial requirements.
