Industrial depth sensing forces trade-offs: precision or range, indoor or outdoor. We tested Onsemi's iToF sensor to see if it breaks these compromises.

Our computer vision team recently completed a hands-on evaluation of the Onsemi Hyperlux ID AF0130, a high-resolution indirect Time-of-Flight sensor designed specifically for industrial automation. The technology promises to address limitations that have kept phase-based depth sensing confined to controlled environments.

Our evaluation covers real-world performance testing, calibration requirements, and integration realities that don’t appear in datasheets. While dual-frequency operation delivers on its core promise, successful deployment demands engineering discipline around calibration, power delivery, and environmental adaptation. This article shows what works, what requires careful attention, and where the Hybrid mode could expand capabilities into outdoor applications.

1. When Cameras Learn to See in Three Dimensions


The factory floor is changing. Walk into a modern warehouse today and you’ll see robots that don’t just scan barcodes, they understand space. They know how tall that pallet is, how far away that obstacle sits, whether that package will fit in the truck. This spatial understanding relies on 3D sensing technology that has been evolving for years, with each generation bringing new capabilities and addressing limitations that prevented widespread industrial adoption.

3D depth sensing isn’t new. Stereo vision systems, structured light projectors, and laser-based range finders have existed for decades. But these technologies have historically forced difficult trade-offs: choose between indoor or outdoor operation, between high resolution or long range, between compact size or mechanical robustness. As autonomous systems demand more from their sensors, navigating complex environments, manipulating diverse objects, operating reliably across varied lighting conditions, these trade-offs have become increasingly problematic.

Side-by-side comparison of 2D camera with flat image capture versus 3D structured light camera projecting blue LED pattern for depth sensing
Traditional 2D camera (left) captures only intensity information without depth data, relying on natural surface features for any depth estimation. 3D structured light camera (right) projects known patterns (blue LED grid) onto the scene, enabling depth measurement even on featureless surfaces by analyzing pattern deformation. Structured light systems require controlled lighting and precise projector-camera alignment. Source

2. Understanding Depth Sensing Technologies


Multiple depth sensing technologies compete in industrial applications, each with well-understood strengths and limitations. While other approaches exist (radar, sonar, ultrasonic), this overview focuses on the dominant optical depth sensing methods in industrial automation. A brief comparison provides context for where the latest generation of Time-of-Flight sensors fits in this landscape.

Stereo Vision

Stereo cameras mimic human vision by using two lenses separated by a fixed distance. Each camera captures the same scene from a slightly different angle, and software analyzes how objects shift between the two views to calculate distance. This approach offers relatively standard camera hardware, keeping costs reasonable, but requires distinct visual features to match between images and suffers accuracy degradation with distance.

Diagram of stereo camera depth calculation using triangulation between two viewpoints with known baseline
Stereo vision uses two cameras at fixed and known separation to capture the same object. Depth is calculated by measuring how much the object shifts (disparity d) between left and right images. Accurate matching requires distinctive surface features, making this approach unreliable on uniform or repetitive textures.
Structured light depth sensing diagram showing projector casting striped pattern onto spherical target with camera sensor observing pattern deformation
Structured light creates artificial texture by projecting patterns (typically infrared stripes or dots) onto surfaces. The camera observes pattern deformation to calculate 3D geometry, eliminating the need for natural surface features. Source

Structured Light

Structured light systems overcome the texture problem by creating their own features. A projector casts a known pattern onto the scene while a camera observes how that pattern deforms across surface geometry to calculate depth.

However, this approach introduces new constraints. Bright ambient light, particularly sunlight, washes out the projected pattern and significantly degrades depth measurement quality. Mechanical alignment between the projector and camera must remain precise, any shift from vibration or mechanical stress reduces accuracy, requiring periodic recalibration to maintain performance.

Time-of-Flight: Direct and Indirect Approaches

Time-of-Flight sensors abandon triangulation entirely, instead measuring how long light takes to travel to an object and back. Because each pixel independently measures distance, these sensors generate dense depth maps without requiring texture or suffering mechanical alignment issues. The compact, monolithic design proves more robust than systems requiring precise baseline maintenance.

Time-of-Flight sensors abandon triangulation entirely, instead measuring how long light takes to travel to an object and back. Because each pixel independently measures distance, these sensors generate dense depth maps without requiring texture or suffering mechanical alignment issues. The compact, monolithic design proves more robust than systems requiring precise baseline maintenance.

Direct Time-of-Flight

Uses laser pulses with precision timing circuits to measure absolute travel time. The timing accuracy required limits pixel density, but these systems excel at long-range outdoor applications. Mechanical scanning mechanisms in traditional lidar add cost and complexity, though solid-state variants are emerging. Autonomous vehicles rely heavily on direct ToF for obstacle detection at extended ranges.

Indirect Time-of-Flight

Modulates emitted light at high frequency and measures phase differences between outgoing and returning waves rather than absolute time. This phase-detection approach enables much denser pixel arrays because the circuits are simpler than precision timing systems, producing high-resolution depth maps suitable for detailed object recognition. However, traditional iToF implementations struggle in bright ambient light, where continuous-wave modulation becomes overwhelmed by sunlight.

Comparison diagram of direct ToF using pulse timing versus indirect ToF using phase comparison between modulated light waves
Direct Time-of-Flight (left) measures the absolute time for light pulses to return using precision timing circuits, enabling long-range detection with lower resolution. Indirect Time-of-Flight (right) uses continuous modulated illumination and measures phase shift between emitted and returned waves, allowing denser pixel arrays for high-resolution depth maps at shorter ranges. Source

3. The Hyperlux ID: Testing Onsemi's Industrial Depth Sensor


We would like to thank EBV Elektronik and Onsemi for providing us with the Hyperlux ID sensor for evaluation and testing. This opportunity allowed us to thoroughly assess the technology’s real-world performance in industrial conditions.

onsemi’s Hyperlux ID family addresses specific limitations that have constrained iToF adoption in demanding industrial applications. The sensor delivers several key advantages:

  • Higher resolution depth mapping: 1280×960 pixel resolution (1.2MP) provides significantly more detailed depth information compared to standard VGA resolution (640×480) ToF sensors, enabling better object detection and finer feature recognition
  • Motion artifact elimination: Global shutter operation captures the entire frame simultaneously, preventing distortion in dynamic scenes such as fast-moving conveyor systems or robotic pick-and-place operations
  • Extended depth range: Dual-frequency modulation architecture balances precision and range, enabling depth measurements up to 30 meters while maintaining sub-centimeter accuracy at closer distances
  • Compact optical format: 1/3.2″ form factor allows integration into space-constrained industrial equipment

The family consists of two variants that differ in their processing architecture and operating modes:

onsemi AF0130 time-of-flight sensor evaluation board with laser drivers mounted on tripod for testing
Onsemi AF0130 evaluation hardware used for testing. The compact form factor demonstrates the practical size advantages of iToF technology over stereo or structured light alternatives.

AF0130

  • Includes onboard depth processing that converts raw phase measurements to calibrated distance values directly on the chip
  • Reduces host system computational load by delivering processed depth maps rather than raw phase data
  • Optimized for applications requiring immediate depth output with minimal host processing

AF0131

  • Removes onboard depth processing, outputting raw phase data that requires host-side processing for depth calculation
  • Includes Hybrid mode sensor feature that combines continuous-wave modulation with pulsed laser emission for operation in high ambient light conditions
  • Optimized for applications requiring outdoor operation or custom depth processing algorithms

Our evaluation focused on the AF0130 variant, testing its dual-frequency architecture and onboard depth processing across typical industrial scenarios.

Side-by-side comparison of industrial pallet depth sensing: left shows monochrome confidence map of a honeycomb-pattern pallet, right displays colorized depth visualization with rainbow thermal mapping showing measurement precision from 0.05m (blue) to 2.00m (red)
This dual-view visualization demonstrates the exceptional depth accuracy of onsemi's Hyperlux ID sensor on a pallet with honeycomb (alveolar) void pattern. The left panel shows the confidence map*, while the right panel presents the depth visualization with thermal color mapping (blue = 0.05m, red = 2.00m). The sensor clearly resolves individual circular voids and structural features despite the geometric complexity.

Technical Architecture: How It Works

The AF0130 operates on the indirect Time-of-Flight principle, where the sensor illuminates the scene with modulated infrared light and measures the phase shift of reflected light at each pixel to calculate distance. This approach enables the 1280×960 pixel array to simultaneously capture depth information across the entire field of view.

The dual-frequency modulation represents the sensor’s key architectural innovation. Lower modulation frequencies extend unambiguous range, while higher frequencies improve depth precision. The sensor can capture both frequencies sequentially in a single frame cycle, delivering both long-range capability and high precision without requiring separate captures or interpolation. The onboard depth processing converts raw phase measurements into calibrated distance values, reducing latency and simplifying integration.

The sensor offers multiple operating modes optimized for different scenarios. Single-frequency modes prioritize either maximum range or maximum precision depending on application requirements. The dual-frequency mode captures both modulation frequencies sequentially, enabling the system to detect objects at extended range while maintaining high precision for nearby features.

Real-World Performance: What The Testing Revealed

Evaluation testing with the AF0130 focused on quantifying real-world performance across the operating conditions typical in industrial automation. The results demonstrated impressive capabilities and practical considerations that integrators should understand for optimal deployment.

Range and Precision

Testing confirmed sub-centimeter precision when properly calibrated, with measurements taken at approximately 1.4 meters distance. The sensor’s modulation frequencies provide different unambiguous range windows: the 75 MHz mode offers approximately 2 meters unambiguous range, while the 100 MHz mode extends this to approximately 7.5 meters. Dual-frequency operation successfully resolved depth aliasing artifacts that would corrupt measurements from objects at mixed distances.

Sunlight resistance test showing intensity map (left) and depth visualization (right). Window glass shows complete measurement failure (black regions in intensity, invalid depth). The visible person's reflection on the window glass demonstrates the sensor's struggle with bright ambient conditions. Indoor areas maintain valid depth measurements.

The Ambient Light Challenge

Testing in an office environment with sunlight through glass windows revealed the expected vulnerability of continuous-wave iToF operation. The confidence rejection algorithm successfully identified pixels contaminated by ambient light, marking them as invalid. Direct sunlight created characteristic black stripes in the confidence map where the beam entered through windows, and window glass itself appeared black in depth measurements. The desk surface in sunlit areas appeared very dark in the intensity map.

The confidence-based rejection prevented corrupted depth measurements from appearing as valid data, a significant achievement given that traditional ToF and iToF sensors typically fail in direct sunlight conditions. The sensor worked reliably out of the box even in these challenging lighting scenarios.

The AF0131’s Hybrid mode sensor feature can further enhance performance by combining pulsed and phased operation for improved ambient light handling, enabling true outdoor operation and extended range capabilities. This hybrid approach provides higher dynamic range through different threshold parameters for various distances, preventing saturation while maintaining measurement accuracy.

The Calibration Reality

The AF0130 exhibited visible vertical striping in depth maps, with groups of columns showing slight but consistent depth bias compared to their neighbors. This behavior underscores that factory defaults serve as starting points, and proper calibration tailored to the specific application is necessary to achieve optimal performance.

The standard calibration process, while documented, requires careful attention to achieve optimal results. Calibration effectiveness varied across different frequency presets, with some configurations responding better than others. Successful deployment requires understanding which operating modes will work reliably for specific applications, making it valuable to work with experienced integrators who understand the nuances of different operating configurations.

Two side-by-side sensor calibration images: left shows a color gradient depth map with visible vertical striping artifacts in bands of red, orange, yellow, green, cyan, and blue; right shows a rainbow-colored curved point cloud against a black background displaying the same vertical banding pattern from the uncalibrated AF0130 depth sensor at 75MHz.
Pre-calibration depth visualization of the AF0130 sensor showing significant vertical striping artifacts. The left panel displays the depth map with distinct color-banded vertical columns indicating non-uniform depth readings across the sensor array. The right panel shows the corresponding point cloud, where the same striping pattern manifests as curved distortions in the 3D reconstruction. Proper calibration eliminates these vertical striping artifacts, reducing residual error to approximately 3mm or less, enabling the sensor to achieve uniform depth measurement across its entire field of view.

4. Strategic Implications: Where This Technology Fits


The AF0130’s capabilities position it for specific applications where its strengths align with real-world requirements and its limitations can be managed or mitigated.

The combination of high resolution and dual-frequency ranging with onboard depth processing makes the sensor particularly well-suited for logistics automation. Volumetric measurement systems that calculate package dimensions for optimal loading benefit from both the resolution and range flexibility. Robotic systems that must navigate while performing precision tasks, detecting pallets several meters away while docking with centimeter-level accuracy, leverage the dual-frequency modes that deliver both capabilities from a single sensor. Mobile robots operating indoors where the primary challenge is obstacle detection in varied lighting rather than outdoor sunlight find a capable solution without the complexity of managing pulsed laser systems.

Manufacturing applications requiring detailed depth information at relatively close range align well with the high-resolution single-frequency modes. Quality control systems inspecting parts for dimensional accuracy, assembly robots verifying component presence and orientation, and packaging systems confirming product placement all benefit from the dense depth maps and consistent accuracy that the AF0130’s onboard depth processing delivers.

Depth sensor applications: warehouse navigation, package scanning, quality inspection, facial security
Key applications leveraging high-resolution depth sensing: autonomous warehouse navigation with real-time obstacle mapping, volumetric package measurement for shipping optimization, precision 3D quality control comparing scanned geometry to reference models, and anti-spoofing facial recognition using genuine depth structure for secure access control.

Biometric security systems, particularly facial recognition that must distinguish between live subjects and photographs or masks, gain genuine depth information that makes spoofing dramatically more difficult. The sensor’s resolution captures sufficient facial detail while the depth information from the AF0130 provides anti-spoofing protection that flat cameras cannot offer.

Applications in bright or outdoor environments would benefit from the AF0131’s Hybrid mode sensor feature, which was not included in this evaluation.

5. Looking Forward


The transition from two-dimensional to three-dimensional sensing enables fundamentally different capabilities in industrial automation. Robots that understand spatial relationships can manipulate objects in unstructured environments. Security systems that measure depth can distinguish genuine subjects from photographs. Automated systems that perceive volume can optimize resource usage in ways that flat images never enabled.

The Onsemi Hyperlux ID sensor family addresses real limitations in industrial depth sensing: the AF0130’s dual-frequency approach with onboard depth processing solves the range-precision trade-off that has constrained single-frequency iToF systems, while the AF0131’s Hybrid mode sensor feature promises to expand iToF capabilities into high-ambient-light environments where continuous-wave systems fail.

However, bridging sensor capability to production deployment requires expertise beyond hardware selection. Successfully integrating the Hyperlux ID demands attention to calibration procedures tailored to specific operating modes and applications, sensor tuning, and understanding which configurations deliver reliable performance for particular use cases. Achieving optimal results requires hands-on experience with the technology’s practical nuances.

6. Conclusion


For organizations ready to invest the engineering effort, the Hyperlux ID provides a solid foundation for spatial intelligence in industrial automation. It bridges the gap between the simplicity of traditional cameras and the rich spatial understanding required by autonomous systems, allowing integrators to approach deployment with both ambition and realism about the work required.

At Deep Vision Consulting, we’ve spent years navigating the gap between sensor datasheets and production-grade deployments. From technology evaluation and calibration strategies to custom integration design, we help organizations deploy 3D vision systems that actually work in real-world industrial environments. Contact us to discuss your project.

The technology is ready. The question is whether the integration discipline matches the silicon’s sophistication.

Related Posts