Entering a world that anticipates, automates: NXP
Over the next decade, Reger predicts that tens of billions of smart, connected robots will be embedded in homes, cars, hospitals and infrastructure.
However different their form factors, they will share four essential capabilities: Sensing, thinking, connecting, and acting.
Devices will perceive their environment, draw on cloud intelligence, make decisions, and execute them physically.
Yet none of this matters, he warned, if users cannot trust the device. Trust means functional safety and cybersecurity – braking systems that never fail, thermostats that do not overheat homes, and cars that cannot be hacked. Autonomous systems have faltered before because, in his view, they were designed incorrectly.
The solution, he argued, is to take inspiration from biology.
The human body separates fast, deterministic reflexes from higher cognition, Not every decision requires a vast AI brain.
“How big does the AI really have to be?” he asked, noting that an ant’s 100,000 neurons can create an efficient transportation system. The future, according to Reger, lies at the edge: Ultra-low-power chips, AI accelerators consuming as little as seven watts, radar that can see through fog and car-to-car communication that functions almost like telepathy.
Data centres will remain important, but the real democratisation of AI, he insisted, “lies at the famous edge in the end device” – in billions of quiet, trusted machines that anticipate our needs before we voice them.
