Targetless calibration for the complete sensor stack, from a recording you already have.

The engine calibrates the complete sensor stack in one joint run: directional and 360° lidar, single and multi-camera rigs, IMU, and INS/GNSS — intrinsic, extrinsic and temporal parameters solved together. Self-serve today: lidar–camera extrinsics in the web app and the API. Multi-lidar, lidar–IMU, lidar–vehicle, radar, camera intrinsics and time offsets run through a Deepen engagement on the same engine. The free Data Checker CLI checks whether a recording is eligible for lidar–camera, multi-lidar, lidar–vehicle and lidar–IMU calibration.

Launch pricing: $150 per calibration (list price $1,500, 90% off until December 31, 2026). Lidar–camera extrinsics for every camera in the web app or through the API; recordings up to 50 GB. Two ways to buy: in the web app you run first, preview free and pay $150 only to download the result; with an API key you pay first and integrate calibration into your pipeline. A failed job is returned to you automatically. Everything else the engine does — multi-lidar, IMU and GNSS in the joint solve, camera intrinsics, time offsets, radar, custom formats, customer-hosted Docker — is a consultative Enterprise engagement, priced per engagement. See pricing.

Yes. Email a link to one recording (.bag, .db3 or .mcap, shared from Google Drive, S3, Dropbox or similar) to calibrate-support@deepen.ai, with rough camera positions if you have them (from CAD, a URDF or /tf_static); otherwise we work them out with you. Our engineers run it and send you the before-and-after preview, usually within a day. If you want the results, it is the same $150 launch price; if not, you pay nothing. Through December 31, 2026, we also set up the first run with any team that would rather do it themselves.

With the hosted API, your recording is uploaded to Deepen's service and used only to run your job. If you need your data to never leave your own infrastructure, the AWS Marketplace path (coming soon) runs the same engine entirely inside your own VPC. For specifics on retention and deletion, contact us.

No. The web app needs a company or university email; personal addresses such as Gmail, Outlook/Hotmail and Yahoo can't run calibrations. Calibrations run on your organisation's recordings and the results are delivered to that organisation, so we tie each run to it.

We report repeatability, not a single blanket accuracy figure: on the same vehicle, independent 30-second recordings agree to within 0.1° angular, 1 ms temporal and 2 cm translation. Absolute accuracy depends on your rig and scene, so every job also returns a confidence score per sensor, quality flags, error stats and a before/after overlay report with the seed-versus-recovered delta, so you can judge the result yourself. Run the free Data Checker CLI first to catch anything likely to produce a low-confidence result.

Deepen has calibrated production sensor rigs for automotive and robotics teams since 2017, including Applied Intuition, Aptiv, BMW, Bosch, Lucid, Nissan, Uber and VW, and the targetless method is patented. Open-source tools calibrate one sensor pair at a time with a target board and stop being maintained when the research project ends; our engine solves the whole rig jointly, from an ordinary recording, and is maintained and supported as a product. An AI model can write a calibration script in an afternoon. It cannot give you a validated engine, the rig-specific edge cases learned across a decade of real vehicles, same-vehicle repeatability to a tenth of a degree, a report that stands up in review, or a team that answers when a result looks wrong. That is what you buy.

You skip the cost of a target rig or calibration bay, a scripted capture session, and a site visit — you calibrate from a recording you already have. Turnaround is 15–25 minutes versus roughly a business day for a typical vendor capture-and-review service, and the launch price is $150 per calibration (list $1,500) with no monthly commitment.