Search papers, labs, and topics across Lattice.
3
0
6
1
Multi-task RL agents solving related navigation tasks underwater rely on a surprisingly small fraction of their weights (1.5%) to differentiate between tasks.
Contextual multi-task reinforcement learning enables autonomous underwater vehicles to nimbly adapt to diverse reef monitoring tasks and turbulent underwater conditions, paving the way for more sustainable marine ecosystem observation.
By predicting semantic changes in underwater video and filtering out ego-motion, DINO-Explorer lets autonomous vehicles focus on real discoveries, not just where they've been.
