Search papers, labs, and topics across Lattice.
6
0
7
Achieving a 14-point boost in grounding accuracy, VistaRef redefines how we approach spatial orientation in AR and human-robot interaction.
PointVG-R achieves a groundbreaking 15.86-point boost in mIoU by integrating geometric reasoning into visual grounding tasks, reshaping our approach to spatial interpretation in models.
Forget monolithic trajectories: PrimitiveVLA teaches robots reusable motion primitives, boosting data efficiency and zero-shot generalization in manipulation tasks.
Achieve autonomous laparoscope control by translating multimodal surgical data into a single "wrench" that guides the robot's movements.
By combining CNNs and State Space Models, DA-Mamba achieves efficient global-local feature alignment for domain adaptive object detection, outperforming prior CNN-only and Transformer-based approaches.
LoRA's feature learning isn't as stable as we thought, but a simple weight-shrinkage trick can fix it, boosting performance without extra memory.
