LiDAR & Depth Sensors for Humanoid Robots: Shipping Hardware, Real Specs, and India Availability

The Current State of Depth Perception in Humanoids

Humanoid robots require reliable three-dimensional perception to navigate unstructured environments, manipulate objects, and maintain dynamic balance. Unlike wheeled or tracked platforms, bipedal systems face continuous vibration, rapid orientation changes, and strict power budgets, which directly influence sensor selection. The industry has moved away from mechanical spinning LiDAR toward solid-state LiDAR, time-of-flight (ToF) cameras, and stereo depth systems. Each technology class offers distinct trade-offs in range, resolution, power consumption, compute load, and environmental robustness. This article evaluates shipping hardware first, pilot deployments second, and public announcements last, providing measurable specs and real-world integration notes for engineers and procurement teams.

Solid-State LiDAR: From Bench to Boots

Solid-state LiDAR eliminates moving mirrors and rotating assemblies, relying on MEMS mirrors, optical phased arrays, or flash illumination to generate point clouds. For humanoids, the primary advantages are shock tolerance, reduced maintenance, and lower rotational inertia. Currently, the most widely shipped units include the Ouster OS0 and OS1 series, Hesai’s Pandar and XT32 lines, and RoboSense’s Helios and Jupiter platforms. These sensors typically deliver 100 to 128 vertical channels, 100 to 200 meters detection range at 10% reflectivity, and point rates exceeding 1 million points per second.

Integration into humanoid torsos and limbs requires careful mounting. MEMS-based units tolerate up to 50 g of vibration without calibration loss, while flash LiDAR offers instant full-frame capture but sacrifices range. Humanoid manufacturers often mount solid-state LiDAR on the head or chest to maintain a stable baseline relative to the center of mass. Power draw ranges from 3 to 8 watts, with I/O handled via Ethernet (1000BASE-T1) or CAN FD. Thermal management remains a constraint; enclosed torso housings require passive heat sinks or low-speed fans to maintain junction temperatures below 85°C during continuous operation.

Time-of-Flight (ToF) Cameras: Range and Resolution Trade-offs

ToF cameras measure depth by calculating the phase shift of modulated infrared light. They produce dense depth maps at high frame rates, making them suitable for close-range manipulation and indoor navigation. Shipped modules include the Intel RealSense D405 and D455, Sony DepthSense series, and ams OSRAM’s SR300 family. Typical specifications cover 3 to 5 meters effective range, 640x480 to 1280x720 resolution, and 30 to 60 fps depth output. Power consumption sits between 1.5 and 4 watts, with USB 3.0 or MIPI CSI-2 interfaces.

The primary limitation is ambient light interference. Direct sunlight or high-intensity indoor lighting can saturate ToF receivers, causing depth dropouts. Manufacturers mitigate this with IR bandpass filters, adaptive modulation frequencies, and multi-frequency phase unwrapping. For humanoids operating in mixed indoor-outdoor conditions, ToF is best deployed as a complementary sensor rather than a primary ranging tool. Compute requirements are minimal on the sensor side, but post-processing for noise filtering and hole-filling typically runs on an onboard GPU or FPGA.

Stereo Depth Systems: Baseline, Compute, and Calibration

Stereo depth cameras calculate disparity between two synchronized image sensors to derive depth. Baseline length directly impacts maximum measurable distance and angular resolution. Shipped systems include the Stereolabs ZED 2i and ZED 4, Intel RealSense L515 (hybrid stereo-structured light), and various custom setups using Sony IMX477 or AR0521 sensors. Baselines range from 12 cm to 30 cm, with effective ranges of 0.5 to 25 meters. Frame rates reach 90 fps at 720p, with raw disparity output requiring real-time rectification and matching.

Calibration drift is the most common failure mode in dynamic humanoid platforms. Thermal expansion, chassis flex, and impact events can shift extrinsic parameters, requiring periodic re-verification or online calibration loops. Compute load is significant; disparity matching typically consumes 2 to 4 TOPS on modern edge GPUs, with additional cycles for point cloud generation and semantic segmentation. Stereo systems excel in texture-rich environments but degrade in low-light or featureless spaces. Many humanoid developers pair stereo with active IR projectors or ToF emitters to maintain depth fidelity across lighting conditions.

Grading Claims: Shipping Hardware vs. Announcements

Public roadmaps often outpace manufacturing reality. We grade perception claims using a strict hierarchy: shipped hardware verified by unit production, pilot deployments confirmed by operational data, and announcements treated as pre-commitment stage. This prevents speculative marketing from skewing technical assessments.

• Shipping Hardware (Verified): Ouster OS1-64, RoboSense Helios, Stereolabs ZED 2i, Intel RealSense D405/D455, and Sony DepthSense modules are mass-produced, spec-sheet compliant, and available through authorized distributors. Independent teardowns and integration guides confirm power, I/O, and calibration stability.
• Pilot Deployments (Operational): Agility Robotics Digit, Figure 01/02, Unitree G1/H1, and Fourier GR-1 utilize commercial off-the-shelf LiDAR and stereo cameras in early production or beta runs. Deployment data confirms 100+ meter LiDAR range and 5-meter stereo depth accuracy in controlled environments. Vibration isolation and thermal routing remain active engineering tasks.
• Announcements (Pre-Commitment): Several startups and OEMs have unveiled custom flash LiDAR, solid-state phased arrays, and multi-spectral depth sensors. Until units pass AEC-Q100 qualification, undergo production line calibration, and appear in bill-of-materials documentation, these remain development-stage claims.

Procurement teams should request factory calibration certificates, MTBF data, and environmental test reports before committing to pilot integration. Sensor fusion architectures must account for latency, frame synchronization, and spatial registration tolerances before deployment.

India Availability and Approximate INR Pricing

India’s robotics procurement landscape relies heavily on imported components, with customs duties, GST, and logistics adding 18% to 32% to landed costs. Below are approximate INR estimates for mainstream depth sensors, flagged as landed cost estimates where applicable. Prices reflect bulk orders (50+ units) and exclude integration services or warranty extensions.

• Solid-State LiDAR: Ouster OS0/OS1 and RoboSense Helios range from ₹65,000 to ₹1,20,000 per unit. Hesai Pandar XT32 sits near ₹90,000 to ₹1,40,000. Landed cost estimates include 18% customs duty, 3% IGST on freight, and port handling. Distributors like Elementz, Robo India, and Techno Electronics handle clearance and provide basic mounting brackets.
• ToF Cameras: Intel RealSense D405/D455 cost ₹12,000 to ₹18,000. Sony DepthSense modules range ₹15,000 to ₹22,000. Landed cost estimates factor in courier duties and local GST. Calibration tools and SDK licenses are typically free, but thermal enclosures add ₹3,000 to ₹5,000.
• Stereo Depth Systems: Stereolabs ZED 2i/4 prices span ₹25,000 to ₹45,000. Custom stereo rigs with matched sensors and baselines range ₹30,000 to ₹60,000. Landed cost estimates include customs, IGST, and calibration verification kits. Local integration support is available through robotics colleges and system integrators in Pune, Bengaluru, and Chennai.

Procurement in India requires advance import documentation, BIS compliance checks for power supplies, and voltage regulation for 230V AC-to-DC sensor power stages. Many Indian labs partner with Shenzhen-based OEMs for direct shipping, bypassing distributor markups but assuming customs risk. Always verify calibration drift specifications and request factory firmware before deployment.

References

• Ouster Sensor Specifications and Integration Guides: https://ouster.com/sensors
• Hesai Technology Official Product Pages: https://www.hesailidar.com/products
• RoboSense Official LiDAR Documentation: https://www.rosenrobotics.com/product
• Stereolabs ZED Camera Specifications and SDK: https://www.stereolabs.com/zed-camera/
• Intel RealSense Depth Camera D400 Series: https://www.intelrealsense.com/depth-cameras/
• Sony Semiconductor DepthSense Camera Family: https://www.sony.com/en/Semiconductor/products/depthsense.html
• ams OSRAM Depth Sensing Solutions: https://ams-osram.com/products/depth-sensing
• Agility Robotics Digit Platform Documentation: https://agilityrobotics.com/digit
• Figure AI Technical Updates and Platform Specs: https://www.figure.ai
• Unitree Robotics G1 and H1 Official Specifications: https://www.unitree.com
• IEEE Spectrum: Humanoid Sensor Integration and Calibration Practices: https://spectrum.ieee.org
• RoboHub: LiDAR and Depth Sensor Benchmarking for Bipedal Platforms: https://www.robohub.org