Patric Jensfelt

GESTO achieves near-ground-truth performance in reasoning about human activities in dynamic scenes, revealing the power of hierarchical memory structures in robotic perception.

Robots can now navigate safely in unpredictable environments by effectively managing multi-object uncertainties with a novel risk-aware control framework.

Hierarchical distillation with temporal occupancy diffusion enables LiDAR models to achieve unprecedented accuracy in 3D perception tasks.

Radar data, often relegated to a learned feature, can unlock significant gains in 3D multi-object tracking robustness, especially when other sensors falter in adverse conditions or at long ranges.

Forget unreliable two-frame supervision: TeFlow unlocks 33% better scene flow estimation by mining temporally consistent motion cues across multiple frames for self-supervised feed-forward models.