underwater robotics teams

Abstract: We present Underwater C$^{3}$-JEPA (cross-view, control-conditioned, context-extended), an object-centric multi-view predictive world model for near-field heavy-load underwater ROV salvage. Without contact sensors, it predicts in latent space how the task-object state evolves through contact interaction and under the hydrodynamic lag of the vehicle, from synchronized multi-view RGB observations and vehicle control signals. C$^{3}$-JEPA encodes multi-camera observations into task-object and context tokens, fuses cross-camera evidence through held-out-view attention, and directly predicts future states conditioned on control. Weak binding anchors the target and gripper at low annotation cost, while SIGReg sharpens the geometric representation. Experiments show that the learned representation transfers substantially more task-relevant information to downstream probes than a reconstruction-free latent baseline, while keeping the predictor lightweight. The resulting predictive interface supports model-predictive-control (MPC) candidate evaluation and imagined-rollout behavior-agent training. Validation on real underwater video shows the same architecture recovering a withheld camera's object state and staying ahead of persistence, so the recipe transfers beyond simulation.

Abstract: Delayed acoustic positioning packets constrain historical navigation states, but a current-time update evaluates them against a mismatched state, whereas exact rewind/replay re-executes the intervening estimator history. This paper introduces compressed delayed-information projection (CDIP), a causal 15-state error-state Kalman filter (ESKF) treatment for delayed-acoustic unmanned underwater vehicle (UUV) navigation. CDIP retains a source-epoch snapshot and the historical-to-current cross-covariance, then projects the delayed source-epoch acoustic correction directly to the current state without full rewind/replay. Exact fixed-lag rewind/replay out-of-sequence-measurement (OOSM) processing serves as a high-fidelity accuracy reference. In 154 usable paired recordings at a fixed 1.5-s acoustic delay without an outage, CDIP reduced mean trajectory-position root-mean-square error (RMSE) from 1.062 m for the baseline to 0.456 m (57.1%). Its 0.456-m mean was 1.03% higher than the 0.451-m replay mean, while its measured mean per-update runtime was 99.2% lower (approximately 127-fold). A separate predeclared sweep across six fixed delays, with 30 paired recordings per delay, and a truth-supported 9-D consistency analysis bound the interpretation. Additional targeted experiments showed near-replay trajectory accuracy across 50-300-s acoustic outages while preserving sub-millisecond update cost. CDIP therefore provides a compact delayed-information treatment with an empirical accuracy-computation trade-off under the evaluated configuration; the evidence does not establish statistical equivalence or non-inferiority relative to replay.

Abstract: Autonomous underwater navigation typically relies on complex and expensive multi-modal sensor suites designed to prioritize global Simultaneous Localization and Mapping (SLAM) accuracy. However, local reactive behaviors such as coarse navigation and obstacle avoidance require only local consistency---a capability that should be feasible using only a Forward-Looking Sonar (FLS), yet remains largely unaddressed, leaving a critical gap in FLS-only local SLAM. Moreover, existing acoustic SLAM frameworks predominantly rely on sparse feature extraction methods that discard substantial portions of the already information-sparse acoustic returns. To overcome these limitations, this work introduces a dense direct registration approach that aligns full acoustic intensity scans to a recursively updated local map. Real-time execution is achieved via an Inverse Compositional Gauss-Newton optimization strategy that minimizes computational overhead. Experimental evaluations show that this dense method yields significant improvements on translation error compared to sparse keypoint baselines, maintaining stable sub-meter tracking precision over wide displacement gaps. Moreover, this approach delivers odometry performance comparable to multi-sensor fusion pipelines (FLS, DVL, and IMU), bypassing expensive payload dependencies in feature-rich environments. We validate real-world applicability through AUV field trials, running the full local SLAM approach onboard an embedded, resource-constrained computer.