Search papers, labs, and topics across Lattice.
X. Meng, P. Hou, and H. Li are with the Thrust of Robotics and Autonomous Systems, Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China. Z. Zhao and J. Civera are with the University of Zaragoza, Zaragoza, Spain. D. Cremers is with the School of Computation, Information and Technology, Technical University of Munich, Munich, Germany. H. Wang is with the School of Automation and IntelligentSensing, Shanghai Jiao Tong University, Shanghai 200240, China (e-mail: wanghesheng @sjtu.edu.cn)
7
0
6
3
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.
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.
Humanoid robots can now nimbly manipulate objects across much larger workspaces thanks to a LiDAR-powered perception system that eliminates the need for constant repositioning.
A practical VLA model, LLaVA-VLA, achieves strong generalization and versatility on a new benchmark, CEBench, while running on consumer-grade GPUs, eliminating the need for costly pre-training.
By "dreaming" plausible scene completions, Dream-SLAM enables robots to navigate dynamic environments more effectively, achieving better localization, mapping, and exploration than existing methods.
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.
Certifiably optimal solutions to 3D vision problems are now within reach, but choosing the right global solver (BnB, CR, or GNC) requires navigating a complex trade-off between optimality, robustness, and scalability.
