Back to project workspace

Project 03 · Physical AI

Physical Intelligence: LiDAR 3D World Model Data Collection

Build a traceable pipeline from LiDAR point-cloud collection and synchronization through labeling, quality control, and a 3D world-model evaluation.

03/ 06
Good for
Good for data and spatial perception
Resources
Public data, a simulator, or approved equipment
Main deliverable
A data card, collection or simulation protocol, point-cloud schema, visualization tool, quality report, and downstream baseline.

Project brief

From scenario and baseline to reproducible evidence.

Build a traceable pipeline from LiDAR point-cloud collection and synchronization through labeling, quality control, and a 3D world-model evaluation.

The handbook starts with a runnable minimum, then adds real constraints. The final presentation should explain the method, results, failures, limitations, and next steps.

Evaluation focus

Use metrics to make trade-offs visible.

01Valid frames and timestamp completeness
02Registration error
03Label quality
04Downstream 3D task metric

Project handbook

Project brief

Choose a clear setting such as a campus path, warehouse, indoor corridor, parking area, or small robot workspace. Produce a data card and support at least one downstream task: 3D detection, segmentation, occupancy, scene flow, trajectory prediction, or reconstruction.

Minimum version

Use at least five timestamped sequences or 100 frames, from a public dataset, simulator, or approved static collection.

  • A coordinate-system and calibration description.
  • A schema for points, intensity, timestamps, poses, and scene labels.
  • Visualization for single frames, accumulated clouds, and trajectories.
  • Quality checks for empty frames, duplicates, drift, and unusual reflections.

Privacy and safety

Do not collect in private places without permission or publish identifiable people, license plates, addresses, or sensitive areas. Follow campus and venue rules, and attach the dataset license, collection scope, and deletion policy.

Resource note: External code, models, datasets, and platforms remain subject to their own licenses, terms of service, and applicable law.