Movidius' Myriad 2 chip and algorithms power DJI Phantom 4, enabling drone's autonomous features like obstacle avoidance
John Kennedy / Silicon Republic :
Context & Ripple Effects
DJI had been building toward an autonomous consumer drone for months: the developer-focused Matrice M100 introduced object avoidance as an SDK feature in mid-2015, and by March 2016 the [[a:865773|Phantom 4 shipped with obstacle detection, human auto-tracking, and a 45 mph top speed at $1400]]. This piece names the enabler underneath those features — Movidius' Myriad 2 chip and algorithms.
For Movidius, the Phantom 4 was the highest-profile validation of a strategy that already spanned Google's Project Tango smartphone platform and would soon extend to a Lenovo VR partnership. Within six months of this story, Intel moved to acquire the company outright, and by 2017 the line had matured into the Myriad X vision processing unit — making this article the moment the consumer-drone market proved the low-power VPU category had mass-market demand.
First-order effects
- DJI gets a consumer drone whose marquee features — obstacle avoidance and subject tracking — run on Movidius silicon rather than cloud connectivity, letting the Phantom 4 fly autonomously offline.
- Movidius converts its Project Tango pedigree into a flagship consumer design win, giving the startup a shipping product to point at alongside its Lenovo VR discussions.
Second-order effects
- Intel, watching vision chips win sockets in drones and VR, moves to buy Movidius outright rather than compete against it in low-power inference hardware.
- Adjacent device makers follow the same playbook: Lenovo taps the same Myriad 2 part for undisclosed VR projects, showing the chip traveling from drones into wearables-adjacent categories.
Third-order effects
- If the pattern holds, autonomous consumer devices consolidate around dedicated low-power vision processors instead of general-purpose CPUs, which is exactly the trajectory Intel's later Myriad X AI-acceleration launch extends.
- Vision-silicon startups become acquisition targets for large semiconductor firms seeking edge-AI capability, shifting the category from independent suppliers to in-house IP inside Intel-class companies.
The trend: Edge AI is migrating from general-purpose processors to dedicated low-power vision chips embedded directly in consumer devices, with drone autonomy as the proof case that pulled Intel into the market.