Explore the data Contact Us
EN

Studio

KPI evaluation
that powers AIR4D (AFI920).

Sensor-performance KPIs are computed and analyzed automatically across every dataset collected. It's the validation backbone that improves the next AIR4D quickly and consistently, and the same infrastructure is scalable for custom client collaborations and strategic partnerships.

Discuss a partnership

01 — Workflow

From collection to dashboard — end-to-end.

From data collection to the KPI dashboard, every stage is automated and standardized to run reproducibly.

SCENARIO TRACK KPI TRACK 1 Data Collection 2 Data Refinement 3 Dataset Publish 4 Data View 5 Discovery 6 Scenario Dashboard 7 KPI Auto-Eval 8 KPI Analysis 9 KPI Dashboard
Shared Pipeline 1

Data Collection

RIDE vehicle real-world driving + multi-sensor capture (see the Drive page)

Shared Pipeline 2

Data Refinement

capture → sync → calibration → format → meta · GT → QA

Shared Pipeline 3

Dataset Publish

Dataset finalized · Dataset DB updated

Scenario Track 4

Data View

Explore the finished dataset in the 3D Viewer (Scene · Clip · Frame)

Scenario Track 5

Discovery & Custom Scenario

Search/explore/filter by ODD and metadata → update scenario sets or build a custom dataset

Scenario Track 6

Scenario Dashboard

Scenario management dashboard — per-ODD/scenario data distribution and custom-dataset status

KPI Track 7

KPI Auto-Evaluation

Automatic computation of AIR4D sensor KPIs — point cloud · tracking, etc.

KPI Track 8

KPI Result Analysis

Detailed analysis by scenario · Clip · Frame · Object · regression comparison · drift detection

KPI Track 9

KPI Dashboard

KPI management dashboard — per-scenario composite KPIs · AIR4D version-over-version score trends · regression monitoring

02 — Features

Four capabilities, working together.

Studio is built from four capabilities — exploring the dataset, quantifying sensor performance, curating scenarios, and monitoring overall trends.

Feature 01

Dataset 3D Viewer

Explore Clip- and Frame-level data with rich metadata in the 3D Viewer. Synchronized Radar · LiDAR · Camera · GT overlay playback.

Feature 02

KPI Auto-Evaluation

Automatically compute AIR4D point-cloud and tracking KPIs, analyzed by scenario · Clip · Frame · Object.

Feature 03

Discovery & Scenario Curation

Search/filter by ODD and metadata to find data and build custom datasets, with per-scenario distribution managed on a dashboard.

Feature 04

KPI Dashboard

Monitor per-scenario composite KPIs, AIR4D version-over-version KPI score trends, and regression trends on one screen.

03 — Data Viewer

Explore every frame, every sensor stream.

Explore the dataset's Clip and Frame units directly in the 3D Viewer. Play Radar · LiDAR · Camera · GT overlay in sync while reviewing rich metadata.

Data Viewer screen

Highlights

  • Scene meta — recording date · vehicle · ODD · start/end · frame count
  • Per-sensor viewer — Camera frames · LiDAR PCD 3D · Radar scan/track
  • Scan timeline — timestamp matching · gap visualization
  • Perception overlay — object · lane · structure · INS · freespace
  • Advanced 4D Radar analysis — Range-Doppler heatmap · beam/NCI heatmap

04 — KPI Analyzer

Quantify radar performance, end-to-end.

Automatically compute and analyze AIR4D sensor-performance KPIs by scenario · Clip · Frame · Object.

Highlights

  • Object detection · tracking · bird's-eye-view (BEV) accuracy KPIs
  • Recall · N_obj · Doppler MAE · Range MAE · per-scan Pass/Fail
  • Regression comparison · drift detection · weak-subdomain flags
  • Comparative analysis between the sensor-under-test (SUT) and the RIDE dataset's Ground Truth
KPI Analyzer screen

05 — Discovery

Search the dataset, curate your own.

Search/filter by ODD and metadata to find data and build custom datasets, with per-scenario distribution managed on a dashboard.

Discovery screen

Highlights

  • Multi-filter search by ODD and metadata (weather · time of day · road type · objects)
  • Clip/Frame-level result preview · instant 3D Viewer linking
  • Build · save · export custom datasets (scenario sets)
  • Per-scenario distribution · ODD coverage-matrix visualization
  • Identify gap scenarios → link to additional-collection requests

06 — KPI Dashboard

Track KPI trends across every AIR4D version.

Monitor per-scenario composite KPIs and AIR4D version-over-version score trends on one screen, tracking regression and drift.

Highlights

  • Per-scenario composite KPI scores — object · tracking · BEV
  • AIR4D version-over-version KPI score trends
  • Regression trends · drift alerts
  • Per-ODD/scenario weak-subdomain heatmap
  • Version comparison · release-gate metrics
KPI Dashboard screen

07 — Collaboration

Where we collaborate.

Collaboration and partnerships built on the RIDE dataset and Studio's KPI infrastructure. Representative forms are below; if you have other needs, we'll explore them together.

Scenario A

AIR4D Performance Tuning

We support software optimization and configuration so AIR4D performs at its best in your operating environment and scenarios — including regional localization. The RIDE dataset and Studio KPI evaluation serve as the validation backbone.

Scenario B

Edge Case Curation

We rapidly curate the edge-case scenarios you need. Where coverage is short, we collect more with the RIDE vehicle and supply a tailored data package.

Scenario C

Sensor Model Fidelity

We provide real-measurement RIDE datasets to simulation-tool vendors to validate sensor-model fidelity and advance sim-based evaluation capabilities.

Scenario D

Custom KPI Development

We define and automate KPI metrics tailored to your perception stack and evaluation requirements, built on the RIDE dataset and Studio's KPI infrastructure.

Let's explore a collaboration.

Partnerships, custom datasets, or a separate effort built on this data and KPI infrastructure — whatever it is, let's talk.

Discuss a partnership Dataset details