RealSense debuts AI depth camera for robots at Automate 2026

RealSense's D585 Pro runs AI on the camera itself. The new depth camera, unveiled at Automate 2026 in RealSense booth 12036, is pitched as a software-defined, edge AI platform designed to grow with developers over time through SDK updates. The company says the device is aimed at humanoids, autonomous mobile robots, collaborative robot arms, industrial robotics and inspection systems, marking a shift from hardware-only perception to programmable perception that can improve via software as the field evolves.

At the heart of the D585 Pro is a proprietary Gen 5 system-on-chip (SoC) that powers an on-device AI inference engine and the full depth and image processing pipeline. RealSense claims the camera delivers more than twice the depth quality of its previous generation, enabling tighter navigation, more precise manipulation, and richer human-robot interaction without burning bandwidth on a host computer. The D585 Pro supports a sub-15 centimeter minimum range at full resolution and offers a generous 120 by 100 degree field of view at 60 frames per second, a combination that broadens both close-quarters interaction and long-range mapping. The product ships with IP65 protection as standard, a nod to expectations in factory floors, outdoor deployments and other harsh environments.

The lineup is framed as a practical step toward autonomous perception on the edge rather than a pure sensing upgrade. The company reports the camera’s edge AI runs directly on the device, reducing reliance on host computing resources and enabling real-time perception-driven tasks in environments with limited connectivity or high latency. Nadav Orbach, RealSense CEO, described the D585 Pro as “the actualization of the Visual Cortex of Physical AI,” underscoring the move from passive sensing to an active, software-upgradable perception platform. RealSense officials highlighted that the D585 Pro is designed to scale through software-defined enhancements, so developers can add new capabilities via SDKs rather than swapping hardware.

The timing is notable: RealSense unveiled the D585 Pro at Automate 2026, with commercial shipping planned for the first quarter of 2027. The company says early responses from leading robotics players reinforce the market’s readiness for a perception platform that blends depth sensing, on-device AI, and software-defined growth. Analysts and engineers watching the launch point to two practical implications for system design. First, edge processing can lower latency and reduce data bottlenecks when robots must react to dynamic scenes, especially in crowded indoor settings where humanoids or cobots operate in close proximity to humans. Second, a software-first path introduces clear tradeoffs: while SDK updates can unlock new capabilities without new hardware, they also place greater emphasis on software compatibility, version management, and the potential need for ongoing calibration as the platform evolves.

Beyond the specs, practitioners will watch how real-world robots integrate the D585 Pro with existing perception stacks, navigation modules and manipulation controllers. The camera’s extended range and robust FOV could simplify sensor fusion in AMRs, while the hardware protection rating makes it a candidate for harsh production floors. However, as with any edge AI product, success will hinge on the ecosystem: how smoothly developers can leverage new SDK features, how updates affect latency and power, and whether software-defined improvements translate into measurable gains in accuracy and reliability across varied use cases.

In short, RealSense is betting that perception is a programmable asset, not a one-off sensor. If the D585 Pro delivers on its edge AI promise and the SDK ecosystem stays coherent across updates, robots that once relied on heavy external processing could become more autonomous, more capable, and easier to maintain in production.