Skip to content
MK
Menu
中Contact
M.K. / FIELD NOTES/Long-form essay series
Long-form essay series

01 — 08 essays

Physical AI Foundations

A systems view of AI entering the physical world, from perception and edge inference to AI cameras, device lifecycle and robotics safety.

Series manifesto

This series treats Physical AI as a long-running thread across models, hardware, sensing, energy, products and governance. It is less about one striking robot demo than the repeated capabilities, limits and responsibilities that real-world systems must handle.

Reading path
01

Foundations

Defining Physical AI, real-world consequences and the shared language for the series.

01
01
What Is Physical AI? Why the Next Wave of AI Will Leave the ScreenGenerative AI transformed how we create and process information. Physical AI goes further: machines begin to perceive, reason and act in the real world, where latency, safety, hardware and failure suddenly matter much more.
02

Perception & Events

How computer vision, video understanding and AI cameras turn perception into actionable events.

02
02
When Computer Vision Starts Understanding Events, Not Just ObjectsComputer vision is moving beyond object detection toward streaming video, event understanding and distributed edge inference. The harder question is who defines what counts as an event and what the system is allowed to do next.
04
AI Cameras Need to Understand Events, Not Just Detect ObjectsThe next AI camera advantage is not simply detecting more objects. It is turning continuous video into a small number of trustworthy events through tracking, rules, edge filtering, VLM review and interoperable metadata.
03

Edge Systems

How latency, connectivity, privacy, power and cost determine where inference should happen.

01
03
Why Edge AI Matters Again: The Real Shift Is Where Inference RunsEdge AI is not replacing cloud AI. As AI moves into devices, cameras, robots and long-running agents, latency, connectivity, privacy, power and cost turn inference placement into a product architecture decision.
04

Product Lifecycle

Placing models, hardware, OTA, security, privacy and end-of-life duties in one product framework.

01
05
When Devices Become AI-Powered, Product Management Becomes Lifecycle ManagementAI-powered devices force product teams to manage hardware and model dependencies, capability limits, power budgets, OTA failure paths, data flows, cybersecurity and end-of-life support.
05

Robotics & Safety

From physical uncertainty and household adoption to the fit between humanoid and specialized robots in human-built environments.

03
06
Why Robotics Is Hard: The Real World Has No API ContractRobotics is difficult not simply because models need to improve, but because perception errors, contact physics, latency, mechanical wear and human behaviour accumulate into uncertainty that the system must manage safely.
07
Why Household Robots Still Haven't Taken OffHousehold robots will not become ordinary because one machine can fold a shirt in a demo. They need to save more time than they add in supervision, recovery, maintenance, privacy and safety burden.
08
When Do Humanoid Robots Make Sense?A human-shaped robot can fit spaces and tools built for people. Whether that benefit justifies two-legged mobility and complex manipulation depends on the task and its alternatives.
← Back to all essays