Horizon Robotics' Sparse4D Perception Framework Accepted by TPAMI
rsasaki0109 · x · 2026-08-12
Open-sourced by Horizon Robotics, Sparse4D is a sparse-based, end-to-end multi-view temporal perception model primarily designed for autonomous driving scenarios.
The framework utilizes an encoder-decoder architecture, taking multi-view images and instance features as inputs. It incorporates recurrent temporal fusion with instances propagated from previous frames to output refined 3D anchor boxes. The team announced that the integrated version of Sparse4D (v1/v2/v3 + lidar) has been accepted by the prestigious journal TPAMI.
More from Embodied
- Exploring Humanoid Robot Training: Learning Unseen Terrain via Synthetic Routes — wandb · 2026-08-12
- UChicago Lab Explores Embodied AI and Haptics for Skill Learning — plopesresearch · 2026-08-12
- EITWatch: Smartwatch Uses Backside Electrodes for Over 90% Hand Gesture Recognition — plopesresearch · 2026-08-12
- Fusion startup FAETON-X hits record 1.27e12 neutrons per shot, a first for commercial companies — dunkhippo33 · 2026-08-12
- Robot Founder Uses His Own Humanoid Robot as Post-Surgery Caregiver — CyberRobooo · 2026-08-12
- Open-Sourced Modular Jaw Upgrade for NormaCore Robot Gripper — philfung · 2026-08-12