Robot models
AUVs, ROVs, and underwater vehicle assets with thruster and body models
OceanScale
The ocean simulator for underwater robotics.
OceanScale accelerates development, testing, validation, and deployment iteration for underwater robots.
Runtime Newton + Warp
Scenario AUV / ROV tasks
Evidence Manifested demo assets
Challenge
Field testing is still the bottleneck
Underwater robots still learn in the most expensive test environment on earth: the real ocean.
Field windows scarce
Hardware risk real
Replayability limited
Virtual Ocean
Simulator state is the linchpin for underwater robot iteration
Agents act, state evolves under ocean physics, and sensors return partial observations. OceanScale owns the state: hydrodynamics, ocean fields, robot interfaces, and evaluation — so renderers and policies share one ocean contract.
Actions and setup
Controls and tasks
Policy actions, docking, station keeping, waypoint following, and replays
Ocean conditions
Currents, visibility, terrain, and disturbances that change state evolution
Simulator state Ocean state runtime
Physics ladder
Fossen 6-DOF, grid flow, SPH, MPM, and sparse-volume storage on Newton + Warp
Ocean field
Current fields, turbidity, terrain, and sensor degradation in the same state
State contract
Pose, twist, ocean context, and provenance shared by sensors and policies
Batched envs
Vectorized environments for policy search on a single GPU
Observations and evidence
Sensor observations
Camera, sonar, DVL, IMU, and pressure streams projected from state
Scenario replays
Configured ocean conditions, task settings, and random seeds
Validation metrics
Success rate, position error, attitude error, and robustness summaries
Evidence
Verified media, manifests, and repeatable scenarios
The public demo uses the checked underwater robot video, poster, and manifest already shipped with the site. The page shows exactly what is proven and avoids unverified sim-to-real claims.
Workflows
Built around expensive underwater robotics loops
The launch site emphasizes workflows where repeatability, scenario variation, and evidence records matter most.
Autonomous docking
Validate approach and terminal behavior under current, turbidity, and localization error.
Station keeping
Measure position, attitude, and operational stability under disturbance.
Waypoint survey
Replay pipeline, seabed structure, and terrain-following scenarios.
Close-range holding
Evaluate pose control near equipment-like geometry.
Limited-observation navigation
Stress navigation policies with low visibility and partial observations.
Configured re-runs
Re-run simulator conditions, task settings, and seeds for the next evaluation pass.
Contact
Build the virtual ocean with us
For underwater robotics teams evaluating faster development, testing, validation, and deployment iteration, contact the OceanScale team. The public source repository opens when the release gate is ready.