Calibrate with Confidence
Multimodal sensor calibration for robotics and physical AI
Physical AI is only as good as the data it learns from. MetriCal multimodal sensor calibration by Tangram Vision delivers accurate and repeatable data so you can deploy and scale fast.
The calibration behind leading perception suites
Robotics, autonomous systems, and physical AI companies around the world use MetriCal
to build, test, and scale with confidence.
One calibration for your whole sensor suite
MetriCal by Tangram Vision replaces slow, manual pairwise calibration with one joint, multimodal optimization across cameras, 3D LiDAR, radar, IMU, and odometry, offering results you can trust on the equipment you already use.
Accurate results, end to end
Get multi-sensor calibration in one joint optimization, so errors don’t compound down a chain.
Recalibrate in minutes, not days
New mount or rig? Quickly rerun and keep moving instead of rebuilding your process.
Detailed, actionable reports with every run
Each MetriCal calibration comes with human- and machine-readable data-quality, precision, and accuracy reports. You get more than a pass/fail. You see why a result happened and exactly what to adjust.
No GPU or perfect targets needed
MetriCal works with the hardware you have and solves for warps and bumps in real-world calibration targets.
Remarkable results from robotics teams
We appreciate the back-and-forth with the Tangram Vision team; they’ve been helpful throughout the calibration development process.
Court Hinricher
Senior Staff Hardware Engineer at Kodiak AI
MetriCal brings trust to physical AI
Great physical AI starts with sensors that agree. MetriCal is the calibration layer of your perception stack that aligns every sensor first, so your perception, sensor fusion, and AI training models start from an unshakable foundation.
Better AI training data
With accurate, consistent calibration from MetriCal, your AI models learn from clean data, preventing errors before they spread to every dataset.
Reliable sensor fusion
When cameras, LiDAR, radar, and IMU agree on where things are, fused data lines up so your detection, localization, and planning hold up in the field.
Faster root-cause diagnosis
MetriCal reports on every run, giving teams a clear view of how each sensor performed. When something’s off, you will know why and exactly what to adjust.
Sensor calibration FAQs
Which sensors does MetriCal support?
MetriCal supports CMOS cameras, IR and thermal cameras, 3D LiDAR, radar, IMU, and odometry, all calibrated together in one joint optimization.
How does the free trial work?
Start a free trial with 15 free credits, no payment info required, and run MetriCal on your own sensor data. You only use credits when a calibration succeeds, so you can test, adjust, and rerun without worrying about wasted runs. When you are ready for more, Tangram offers flexible and transparent usage-based and subscription pricing so you can select the plan that works best for your team.
Does MetriCal need a GPU?
No, MetriCal can run on standard computers and edge devices, so it doesn’t compete for the GPU resources your perception and AI models depend on.
How is MetriCal different from other calibration solutions?
In one pass, MetriCal solves intrinsic and extrinsic parameters across your entire multimodal sensor suite, including cameras, LiDAR, radar, IMU, and odometry, so it scales cleanly as your suite grows and uncertainty does not compound down a calibration chain. Every run includes detailed reports on data quality, precision, and accuracy, with the full methodology behind each metric, so you see why a result passed or failed and exactly what to adjust. MetriCal also runs without a GPU, works with imperfect calibration targets, and plugs into your existing tools through MCAP input and OpenCV, ROS, or URDF outputs.
What types of teams use MetriCal?
MetriCal is built for companies whose products rely on accurate, reliable sensor data. That includes teams developing autonomous vehicles and trucks, ADAS, warehouse and delivery robots, drones, autonomous boats, and even lunar rovers. Manufacturers also use MetriCal to power machine vision for visual inspection, quality control, and vision-guided robotics. Whether you’re calibrating your first prototype or scaling production, MetriCal helps your computer vision team get reliable perception into the field faster.
How does calibration uncertainty affect sensor fusion and physical AI?
Calibration errors misalign the data your sensors share. MetriCal offers joint sensor calibration and reports on data quality, precision, and accuracy on every run. When something’s off, you see exactly why and what to adjust, so you can catch issues before deployment instead of debugging them in the field. The result is reliable and repeatable data for sensor fusion and training physical AI your team can trust, with less engineering time spent chasing calibration problems.
Tangram Vision Blog
Explore expert insights from our team on perception, physical AI, and the latest trends shaping robotics and autonomy.
Stop doubting your calibration. Try MetriCal today.
Start with 15 free credits, no payment info required, and turn your sensor data into calibrations you can trust.
