Leopard Imaging (Fremont, CA, USA) and Lumotive (Fremont, CA, USA) have developed a long range RGB+depth perception platform designed especially for next generation machine vision and AI perception applications. The platform, Sirius Max, high-performance imaging, intelligent indirect time of flight depth sensing and processing, embedded synchronization, and programmable bean-steered illumination into a fully integrated GMSL2 camera.

Vision Systems Design wanted to learn more, so we reached out to Dr. Apurva Jain, senior vice president of product, marketing, and customer success at Lumotive for more information.

Editor’s note: The following Q&A may have been edited for style and clarity. 

Vision Systems Design (VSD): What advantages does combining RGB, iToF, and programmable beam steering provide compared with conventional RGB-D and LiDAR?

Apurva Jain (AJ): Conventional RGB-D cameras either flood the whole scene with flash illumination or rely on stereo, which struggles at range and on low-texture surfaces. Mechanical LiDAR adds cost, size, and moving parts. As a result, a typical robot today carries several fixed-function sensors (short-range 3D cameras, a 2D laser scanner, cliff sensors) that add size, complexity, and processing load. Sirius Max combines a global-shutter RGB sensor, the onsemi AF0130 iToF depth sensor, and Lumotive's Light Control Metasurface transmitter. 

The LCM is a solid-state semiconductor chip that steers and shapes laser light electronically, so illumination is concentrated where sensing is needed. That improves optical efficiency, ambient light rejection, signal-to-noise ratio and range, with no moving parts. Synchronized RGB, depth, and control run over a single GMSL2 coax cable, so one module can replace several sensors and reduce system cost and integration effort. 

VSD: How can developers leverage software-defined beam steering and programmable sensing regions?

AJ: Steering angle, beam divergence, dwell time, and scan pattern can all be reconfigured in software, in real time, and presets such as AMR mode tune the module for a given task. For navigation, a robot can scan the far field for obstacles while dedicating dense coverage to the floor ahead for cliff detection. 

For inspection and object handling, it can concentrate resolution on a region of interest at close range. On-the-fly HDR identifies saturated pixels and re-exposes them within the same frame, so both dark and bright objects are captured. 

Because the sensing is programmable, a single platform can serve SLAM, collision avoidance, cliff detection, visual inspection, and navigation, and machines can be tuned to see better over time as the AI stack learns what information it needs. 

VSD: What enabled iToF sensing beyond 20 meters, and which applications benefit most?

AJ: Flash iToF spreads the laser power over the entire field of view, so very little light returns from distant objects. The LCM instead illuminates the field of view in discrete regions of interest and stitches them into a full point cloud, putting far more optical power on each region. That higher signal-to-noise ratio extends range beyond 20 meters indoors and outdoors, while retaining short-range accuracy. 

The biggest beneficiaries are warehouse and factory AMRs and AGVs, where longer range vision enables safer high-speed operation, and humanoid robots, which need both close-range accuracy for manipulation and longer range for navigation from one compact sensor. Outdoor and semi-outdoor robotics and smart infrastructure also benefit.

VSD: How does Sirius Max address ambient light, reflective surfaces, and multipath, and what have you seen in testing?

AJ: Directional illumination concentrates the signal, which improves ambient light rejection outdoors and in bright facilities. Multipath in flash systems comes from lighting the whole scene at once, so light bouncing off walls and floors corrupts the depth measurement, especially near corners. Illuminating one region at a time sharply reduces that interference and removes the blind spots it creates. 

For reflective and high-contrast surfaces, on-the-fly HDR reduces blooming and saturation. Steering also minimizes interference between neighboring units in multi-robot fleets. 

In side-by-side testing against flash iToF and active stereo, LCM-based iToF produces point clouds with less noise, fewer artifacts and blind spots, better near-range precision and more reliable detection of people, pallets, floors and corners. 

VSD: Where does programmable optical sensing fit in the future of autonomous machines and spatial intelligence?

AJ: Robotics is moving from a few robots doing fixed tasks in controlled environments to hundreds of robots per facility working alongside people and adapting to new tasks. Fixed sensors with static performance cannot keep up with that. Programmable optics lets light be controlled as flexibly as software, so the sensor becomes a programmable eye that the AI stack can direct to the information it needs. That enables context-aware sensing: the robot adapts its field of view, resolution, range and frame rate to the task and environment in the moment, for example looking far ahead while moving at speed through an open aisle, then concentrating on the floor and nearby people in a tight space or during manipulation. 

It also enables sensor consolidation, doing more with less. One programmable sensor takes on the job of several fixed-function sensors, which reduces the area and power the perception system needs on the robot. That matters most with humanoids, drones, and compact AMRs, where space and battery budgets are tightly constrained. 

We see programmable optical sensing becoming a core building block for physical AI across humanoids, AMRs, industrial automation and smart infrastructure.

VSD: Can you share a customer use case or pilot deployment that highlights the system’s unique capabilities and return on investment?

AJ: Lumotive's software-defined sensing approach has been proven with deployments in robotic platforms with major OEMs, demonstrating significantly reduced total cost of ownership. Examples include the need for only one sensor instead of two or three; solid-state reliability with less maintenance and more uptime; lower power consumption for less heat and longer battery life; higher frame rates for greater speed and throughput; and greater precision, allowing robots to work closer to walls, people, and objects with fewer false stops.

Related: Focus on Vision : Advancements in Industrial and 3D Technologies

Related: From Prototype to Production: Engineering Edge AI for the Factory Floor