Onboard Optimization and Learning: A Survey

TL;DR

This survey addresses the challenge of enabling robust, low-latency onboard AI on resource-constrained devices by systematically organizing methods into five core pillars: model compression, efficient inference, decentralized learning, security/privacy, and advanced hardware-software co-design topics. It synthesizes techniques such as pruning, quantization, knowledge distillation, NAS, offloading, model partitioning, federated/split learning, and adaptive privacy mechanisms, highlighting how these approaches interact under constraints of latency, energy, memory, and bandwidth. The work provides a unified, cross-layer perspective, emphasizes system-level trade-offs, and discusses real-world deployment risks, benchmarking needs, and future directions toward unified onboard intelligence. By connecting algorithmic improvements with hardware realities and deployment considerations, the survey aims to guide practitioners and researchers toward practical, scalable, and secure edge AI solutions.

Abstract

Onboard learning is a transformative approach in edge AI, enabling real-time data processing, decision-making, and adaptive model training directly on resource-constrained devices without relying on centralized servers. This paradigm is crucial for applications demanding low latency, enhanced privacy, and energy efficiency. However, onboard learning faces challenges such as limited computational resources, high inference costs, and security vulnerabilities. This survey explores a comprehensive range of methodologies that address these challenges, focusing on techniques that optimize model efficiency, accelerate inference, and support collaborative learning across distributed devices. Approaches for reducing model complexity, improving inference speed, and ensuring privacy-preserving computation are examined alongside emerging strategies that enhance scalability and adaptability in dynamic environments. By bridging advancements in hardware-software co-design, model compression, and decentralized learning, this survey provides insights into the current state of onboard learning to enable robust, efficient, and secure AI deployment at the edge.