
November 23–27, 2026
Independent Researcher
Embedded Linux drivers make up a major portion of kernel source code, yet GPIO and pin-control subsystems generally assume exclusive ownership of hardware resources. Shared reset lines, dual-purpose interrupt and status GPIOs, and multiplexed protocol pins frequently violate this assumption, creating integration challenges during silicon bring-up. This talk presents an AI-assisted software-defined resource arbitration framework that enables intelligent coordination of shared GPIO and pin resources while preserving compatibility with existing kernel interfaces. The architecture combines a priority-based access scheduler, state-aware configuration manager, transparent coordination mechanism, and device-tree-defined sharing policies. AI-driven decision logic can analyze runtime resource states, access patterns, and contention conditions to support adaptive allocation and configuration decisions. Hierarchical PID controllers operating across multiple timescales further manage sharing-context transitions, while conditional resource activation and reference-counted dependency tracking maintain system stability without modifications to the kernel core. The framework is validated on resource-constrained IoT hardware using fault-injection techniques to evaluate robustness and runtime behavior. Attendees will learn how AI-assisted resource arbitration can enable dynamic pin sharing, reduce hardware constraints, improve runtime adaptability, and maintain real-time predictability across automotive, industrial, and IoT embedded platforms.
See every session across the AI Agents Conference November 2026 line-up.
View Agenda