Haoang Li

Achieving 98.0% success in cross-embodiment manipulation without manual action alignment could redefine how we approach robot control across diverse platforms.

Robust-WAM achieves superior out-of-distribution generalization in robot control by seamlessly integrating semantic foresight into action predictions while leveraging extensive VGM pretraining.

MobileWAM achieves superior mobile manipulation performance by seamlessly integrating foresight into action planning, outpacing state-of-the-art methods.

UAV swarms can now achieve unprecedented tracking accuracy in complex environments, reducing errors by up to 15% while flying in real-time.

Drones can navigate uncertain winds and complete deliveries more reliably by dynamically re-routing based on real-time energy expenditure and risk assessment.

Robots can now nimbly navigate complex, multi-floor environments without prior training, thanks to a new strategy that dynamically switches between exploration, recovery, and memory recall.

VLN agents can navigate more effectively by predicting their future states and proactively planning based on forecasted semantic map cues, rather than relying solely on historical context.

Ditch slow, multi-step video generation: S-VAM distills the structured generative priors of multi-step denoising into a single forward pass for real-time robot action prediction.

Achieve higher accuracy in dual-arm robot calibration by unifying coordinate and kinematic parameter estimation within a single Lie-algebraic framework, eliminating artificial error separation.

Achieve state-of-the-art outlier robustness and threshold resilience in geometric estimation with a surprisingly efficient branch-and-bound approach.

By aligning latent representations with multiple visual foundation models, FRAPPE offers a more scalable and data-efficient way to imbue generalist robotic policies with robust world-awareness.