Building the systems behind the next generation of intelligent robots

The global industrial robotics market reached a record $16.7 billion in 2024, driven by AI advances that expand robots’ perception and interaction capabilities. Robotics firms now face the challenge of transitioning from experimental prototypes to commercial systems that operate safely, reliably, and economically in dynamic environments. Federico Pecora of Arm argues in a new technical paper that real-world deployment depends on integrating AI inference, sensing, planning, control, safety, and orchestration into cohesive machine systems.

A robot functions as a continuously operating autonomous system, combining inputs from cameras, lidar, radar, and force sensors with AI inference, real-time planning, and safety monitoring. These workloads operate at different speeds and require concurrent execution, creating engineering challenges in data sharing, scheduling, and predictable timing. Motor control demands strict real-time guarantees, while perception and AI inference require high-throughput compute, necessitating careful system design to avoid bottlenecks.

As robots take on more complex tasks, engineering teams must balance performance gains from larger AI models against increased demands on memory, bandwidth, and runtime control. Workload placement—deciding which functions run on CPUs, accelerators, or real-time domains—becomes a critical design choice. CPUs often coordinate sensors, manage data flows, and maintain safety-critical operations even as AI workloads intensify, underscoring their central role in robotic systems.

Arm aims to reduce complexity in next-generation robotics by providing a compute foundation that supports heterogeneous, power-efficient, and scalable architectures. The company’s technical paper examines how to integrate AI capabilities into dependable physical systems, enabling robots to sense, reason, and act safely in real-world settings. By supporting diverse workloads from high-performance processing to low-power sensing, Arm helps engineering teams build robots that balance intelligence, efficiency, and reliability at scale.