RealSense D585 Pro Extends depth to 10 m

A single depth camera now sees from under 15 centimeters to over 10 meters. RealSense unveiled the AI-native D585 Pro depth camera at Automate 2026 in booth 12036, pitching it as a software-defined leap for humanoids, autonomous mobile robots, collaborative arms and industrial inspection systems.

The D585 Pro marries depth sensing with edge AI acceleration on a platform described by the company as software-defined. RealSense says the camera runs a proprietary Gen 5 system-on-chip and an on-device AI inference engine that processes the full depth and image processing pipeline at the edge, reducing the need to ferry data to off-board compute. In practical terms, that means a robot can navigate, manipulate and inspect with less reliance on a workstation or cloud link, potentially cutting latency and simplifying system architectures. The camera is designed for environments that demand ruggedness and reliability, with IP65 protection standard and a field of view of 120 by 100 degrees at 60 frames per second, and a minimum range of under 15 centimeters at full resolution. RealSense contends the D585 Pro delivers more than twice the depth quality of its predecessor, underscoring a push to better precision in cluttered environments, where misreads can derail manipulation or safe human-robot interaction.

The product is pitched for a broad robotics audience, including humanoids, autonomous mobile robots, collaborative robot arms, industrial robots and inspection systems. The D585 Pro is positioned as more than a hardware refresh; the company claims it is the first generation to actualize a software-defined, AI-native perception stack that can evolve over time via SDK-delivered updates rather than wholesale hardware changes. Nadav Orbach, RealSense's CEO, described the device as the “actualization of the Visual Cortex of Physical AI,” emphasizing the promise of a single camera covering a wide depth range indoors or outdoors while gaining new capabilities through software updates.

From a deployment perspective, RealSense notes the D585 Pro is designed to integrate with existing robot platforms that rely on visual inputs to localize, map and grasp objects. The on-device AI processing reduces data traffic to host computers and networks, which can simplify system design and lower bandwidth requirements on production lines. Early responses from leading robotics companies have reinforced the belief that the industry is ready for a new generation of perception platforms that combine depth sensing, AI and software-defined capabilities in a single package. The company says the D585 Pro will begin shipping in Q1 2027, a timeline that places it firmly in the next wave of production-leaning robot architectures rather than a lab prototype.

Two to four practitioner insights emerge from the launch. First, the edge AI on-camera approach promises lower latency and tighter closed-loop control for precise manipulation, but it also tightens the design constraints around power, thermal management and enclosure integration in compact robots. Second, the software-defined angle is a double-edged sword: developers gain rapid feature rollout and customization, yet production deployments must plan for SDK maturity, backward compatibility and long-term support to avoid mismatches between software updates and robot firmware. Third, the 15 cm to 10 m range and the ultra-wide yet high-frame-rate FOV raise expectations for perception reliability in diverse lighting and surface conditions; engineers will want to study how well the system handles reflective materials, sparse features and outdoor glare. Fourth, the shipping timeline in Q1 2027 makes this a live criteria for procurement planning, supplier risk, and field-readiness tests, as integrators gauge whether software-defined advances can outpace competing perception stacks.

The D585 Pro marks a shift toward perception platforms that treat depth sensing, AI inference and software as a single evolving stack. If the early field feedback translates into robust SDKs, clear calibration workflows and solid long-term support, the camera could become a standard building block for next-generation robots that must perceive, reason and act with minimal host dependence.