How to Make a Robot Better at Its Job? Give It Eyes.

“Inbolt built its system around an off-the-shelf industrial camera made by RealSense [A] that costs less than $600. Inbolt’s proprietary software is what differentiates Inbolt from competitors. Its software gives robots the ability to map the environment around them in 3-D and uses CAD models to train robots to recognize parts.

 

“They were the first ones to really crack the code,” said Mike Nielsen, chief marketing officer of RealSense, a start-up incubated at Intel that became a stand-alone company last year. The cameras have high-precision lenses, an infrared projector and sensors that can calculate distance.

 

“I think in the next three to four years, robots that didn’t have cameras before will have them,” Mr. Nieslen said.

 

Inbolt, which sells robot “eyes” as a subscription service, deployed its first system at a Stellantis plant in Hungary in 2024. Since then, it has been installed on over 200 robots around the world. The company has raised roughly $30 million since its inception and employs 50 people around the world, including 15 in the United States.” [B]

 

 

 

 

 

 

 

A. RealSense, a pioneer in 3D computer vision and depth-sensing technology, operates as an independent company following its strategic spinout from Intel Corporation.

Historically known as Intel RealSense, the business secured a $50 million Series A funding round led by a semiconductor-focused private equity firm alongside strategic backers like Intel Capital and the MediaTek Innovation Fund. Headquartered in Cupertino, California, RealSense has shifted its focus heavily toward building the "visual cortex of Physical AI"—primarily supplying specialized computer vision hardware and software for humanoid robots, autonomous mobile robots (AMRs), industrial automation, and biometric access control.

________________________________________

Core Hardware Lineup

RealSense technology relies heavily on stereoscopic 3D cameras and active infrared technology to calculate depth maps in real time.

 

1. Flagship AI-Native Series

•           RealSense D585 Pro: Built on a next-generation hardware architecture, this is the brand's premiere AI-native platform designed for humanoids and complex robotic tasks. It utilizes an on-camera Gen 5 System-on-Chip (SoC) featuring dedicated AI accelerators, a 120°×100° wide field-of-view, a 10cm minimum close-range distance, and simultaneous 30 FPS RGB/depth edge-processing without drawing host CPU overhead.

2. D400 Standard Series (USB)

 

 

D435i

$589.99

category.yahboom.net

 

D435 / D435i: The most widely recognized peripheral form-factors. The D435i features a global shutter for fast-moving environments and an integrated Inertial Measurement Unit (IMU) for basic spatial tracking.

________________________________________

 

 

D415

$400.00

Robotopian& more

3.8

(21)

 

Offers a narrower field of view but higher pixel density per degree, yielding precise meshes and tight dimensional accuracy for static, fine-detail inspection.

________________________________________

 

 

D405

$288.75

DigiKey& more

5.0

(1)

 

A specialized micro-camera optimized for sub-millimeter precision at ultra-close ranges (7 cm to 50 cm), often deployed in surgical medical equipment or close-range industrial inspection.

________________________________________

3. Industrial & Ruggedized Portfolio

•           D457 GMSL/FAKRA: Replaces traditional USB infrastructure with high-bandwidth Gigabit Multimedia Serial Link (GMSL) technology, facilitating data transmission over distances up to 15 meters via robust FAKRA cables. It is built with a dust- and water-resistant IP65-rated enclosure.

•           D555 PoE: A network-integrated camera optimized for physical installations that routes power and data through a single Power over Ethernet (PoE) connection. ________________________________________

Software & Strategic Ecosystem

Instead of operating strictly as a hardware manufacturer, RealSense develops an end-to-end perception ecosystem:

•           Perception Studio & SDK 2.0: An open-source, flexible software development kit that handles complex spatial intelligence, localized mapping (Visual SLAM), 3D collision avoidance, and multi-camera sensor alignment.

•           NVIDIA Integration: RealSense works closely with NVIDIA's robotics ecosystem, ensuring its cameras seamlessly interface with the NVIDIA Jetson Thor platform, Isaac Sim simulation engines, and cuVSLAM for accelerated humanoid navigation.

•           Biometrics & Security: Beyond robotics, the brand markets the RealSense ID Pro, an edge-based facial authentication peripheral utilized by infrastructure networks like airports and data centers for anti-spoofing and secure access control.

 

 

 

B. How to Make a Robot Better at Its Job? Give It Eyes. Stockman, Farah; Rice, Sarah.  New York Times (Online) New York Times Company. Aug 17, 2026.