Leopard Imaging and Lumotive have introduced Sirius Max, a solid-state 3D perception platform that combines RGB imaging, iToF depth sensing and programmable beam steering for robotics and edge AI systems. The GMSL2 camera platform is designed to give autonomous machines software-controlled depth scanning and real-time RGB-depth fusion, with the companies targeting applications ranging from humanoid robotics and industrial automation to autonomous navigation.
AI systems operating in the physical world need more than increasingly capable models. They also need sensors capable of collecting reliable information about objects, distances and environments in real time.
Leopard Imaging and programmable-optics company Lumotive are addressing that perception layer with Sirius Max, a solid-state 3D LiDAR and RGB-depth platform designed for robotics, autonomous machines, industrial automation and intelligent edge systems.
The system combines Leopard Imaging’s embedded imaging and AI-perception capabilities with Lumotive’s programmable beam-steering technology. The companies say Sirius Max can provide long-range active depth perception while allowing developers to control where the sensor directs its illumination and sensing.
That software-defined approach is one of the product’s key differences from conventional time-of-flight systems.
Programmable sensing replaces fixed illumination
Traditional flood-illumination ToF systems broadly illuminate a scene before measuring the returning signal. Sirius Max instead uses Lumotive’s programmable beam-steering technology to direct illumination toward selected areas.
The companies say this approach improves optical efficiency, ambient-light rejection and signal-to-noise performance while avoiding mechanical scanning components. Sirius Max uses Lumotive’s LCM programmable beam-steering module together with an onsemi AF0130 depth sensor.
For AI perception workloads, the ability to control the sensing region in software can be significant. A robot navigating a warehouse, for example, may need detailed depth information in front of its path while requiring less sensing elsewhere. An industrial inspection system may similarly prioritize a particular object or area.
Rather than requiring a different physical sensor configuration for each task, programmable scanning can allow the same hardware to adapt through software.
RGB and depth in one perception system
Sirius Max integrates synchronized RGB imaging, iToF depth sensing, embedded processing and synchronization into a single GMSL2 camera platform.
The system can transmit RGB video, depth information and control signals over a single coaxial connection, according to Leopard Imaging. That architecture is designed to simplify connections between the sensor and an embedded computing platform handling AI inference.
The platform provides what the companies describe as real-time RGB-depth fusion, combining conventional visual information with spatial measurements.
For robotics, that combination can support tasks where recognizing an object is not enough. A machine also needs to understand where the object is located and how far away it is.
Potential applications include simultaneous localization and mapping (SLAM), collision avoidance, cliff detection, visual inspection, object handling and navigation.
More than 20 meters of depth perception
Lumotive and Leopard Imaging say Sirius Max extends iToF sensing performance beyond 20 meters in indoor and outdoor environments and provides 1-megapixel depth resolution. Those figures are manufacturer-provided specifications rather than independent benchmark results.
Longer-range perception is particularly relevant for autonomous machines operating at speed. A robot or vehicle needs sufficient distance to identify obstacles and plan its next movement before reaching them.
For edge AI systems, however, sensor range is only one part of the equation. Latency, synchronization, compute requirements, environmental conditions and the quality of the resulting depth data all influence how effectively perception models can use the sensor.
Sirius Max is therefore positioned as an integrated sensing platform rather than simply a standalone LiDAR component.
Edge AI moves closer to the physical world
The launch reflects a broader evolution in AI infrastructure: machine learning systems are increasingly being designed around physical environments rather than only digital data.
Robotics, autonomous vehicles, industrial machines and smart infrastructure all require a combination of sensors, embedded compute and AI models. This is driving demand for edge AI architectures capable of processing information locally and responding with low latency.
Sirius Max’s GMSL2 connectivity and embedded processing are intended to fit into that architecture. By combining multiple sensing functions into one camera module, the companies argue that developers can reduce the number of components required for perception systems.
That can be particularly important in robotics, where weight, power consumption, physical space and cable complexity can directly affect system design.
Software-defined optics changes the sensor equation
The larger significance of Sirius Max is its combination of programmable optics and AI perception.
AI models can change frequently as developers improve perception, navigation or object-recognition software. A sensor with software-defined scanning can potentially change its sensing behavior alongside those applications.
This creates a more flexible interface between physical sensing and machine-learning infrastructure. Instead of treating the sensor as a fixed input device, developers can use software to determine where and how the environment should be measured.
Lumotive has positioned its Light Control Metasurface technology around programmable control of light, turning optical functions into electronically controlled operations. Sirius Max applies that approach to a combined RGB-depth perception system.
For humanoid robots and autonomous machines, this could support increasingly adaptive perception pipelines in which sensing priorities change according to the machine’s task and surroundings.
The immediate market remains focused on specialized applications rather than general-purpose AI. But as physical AI expands, perception hardware becomes an increasingly important part of the overall AI stack. Sirius Max illustrates that shift by combining imaging, depth sensing, programmable optics and embedded connectivity into a platform designed specifically for machine perception.
Market Landscape
AI perception is becoming a critical infrastructure layer for robotics, autonomous systems and industrial automation. Computer vision provides semantic information, while depth and LiDAR systems add spatial understanding required for navigation, manipulation and obstacle detection.
The market includes conventional LiDAR manufacturers, camera suppliers, 3D ToF companies and newer programmable-optics providers. The trend toward software-defined sensing adds another competitive dimension, allowing manufacturers to differentiate through how sensors can be configured and integrated with AI workloads.
Sirius Max sits at the intersection of these markets, combining RGB imaging and iToF depth with programmable beam steering and edge connectivity. Its stated range exceeds 20 meters, while its software-defined scanning is designed to let developers configure sensing regions for specific applications.
The broader technology trend is toward tightly integrated perception stacks in which sensors, embedded processors and AI models operate as a coordinated system rather than as separate components.
Top Insights
- Sirius Max combines RGB imaging, iToF depth sensing and programmable beam steering into a single GMSL2 platform for AI perception.
- Software-defined scanning allows developers to control sensing regions, potentially adapting perception behavior to different robotic and industrial workloads.
- Leopard Imaging and Lumotive claim more than 20 meters of iToF sensing performance and 1MP depth resolution across indoor and outdoor environments.
- The platform targets physical AI applications including SLAM, collision avoidance, object handling, navigation and industrial inspection.
- Programmable optics could make perception hardware more adaptable as AI models and autonomous-machine requirements evolve.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
