Search papers, labs, and topics across Lattice.
6
0
10
Solving multi-camera relative pose estimation with just four point correspondences could revolutionize the efficiency of visual odometry in autonomous systems.
Achieving real-time relative pose estimation in autonomous vehicles with minimal computational overhead could redefine the efficiency of navigation systems.
Semantic similarity is a poor proxy for retrieval when task success hinges on applying the right rules, so ditch it and retrieve based on applicability.
Robots can now learn to "handle with care," reducing grip force by over 70% in gentle manipulation tasks, thanks to a new vision-tactile-language model and benchmark.
Instruction-following in large reasoning models gets a serious upgrade with RAIN-Merging, a gradient-free technique that merges in instruction-tuned capabilities without wrecking the model's ability to think step-by-step.
Achieve zero-collision embedding tables in production recommenders without sacrificing training speed, unlocking better personalization via fresher and higher-quality item embeddings.
