Abhinav Valada

By fusing confidence-weighted point cloud projections with a Kalman-inspired update mechanism, ConfCtrl enables diffusion models to generate geometrically consistent novel views from sparse inputs, even under significant viewpoint shifts.

Autonomous vehicles can now better adapt to the messy, ever-changing real world thanks to a new motion forecasting method that learns new object classes on the fly without forgetting old ones.

Forget explicit labels: this method learns object co-occurrence priors directly from unlabeled visual data, rivaling human search efficiency.

BLINK unlocks a unified framework for quantitative evaluation and structured modeling of NK cytotoxic behavior at the single-cell level, moving beyond static frame-wise analysis.

By embedding interpretable, sparse $\ell_1$-regularized regression within a neural network, this approach unlocks the ability to extract and understand the key drivers of temporal dynamics in complex cell imaging data.

Autonomous driving gets a boost with LAD-Drive, a new method that uses probabilistic meta-actions and diffusion to generate safer, more nuanced trajectories, outperforming existing methods by a significant margin.

Achieve state-of-the-art 4D scene understanding by representing dynamic environments as a collection of Gaussians that are efficiently splatted onto a voxel grid for panoptic occupancy tracking.

Robots can now build 3D scene graphs that understand how objects move, enabling more robust manipulation of articulated objects in real-world environments.

Skip directly learning actions and instead infer the "what" of robotic manipulation tasks by extracting sparse graphical task models from just a few demonstrations.