Evolution of Robotic Grasping: Cumulative Progress from Dex-Net to GraspGen
DJiafei · x · 2026-08-23
While teaching a new class and reviewing foundational papers, the author highlights a recurring lesson in robotic grasping: progress is often cumulative, not revolutionary.
- Core Insight: Better models, data, compute, and tools enable enduring ideas to move forward.
- Historical Context: This trajectory spans from optimal grasp planning to key works like Dex-Net, Contact-GraspNet, and GraspGen.
Related event: Revisiting Classic Papers: Robotic Grasping Advances Cumulatively(2 posts)→
More from Embodied
- Humanoid robotics shift toward end-to-end control for direct perception-driven movement — aftahi_ai · 2026-08-23
- Bionic hand smartARM pushes dexterity even further — dunkhippo33 · 2026-08-23
- Hands-on with Samsung Odyssey 3D: Comfortable Parallax but Limited FOV — Darpinian · 2026-08-23
- Stanford/PKU paper: never train inside the world model — QWM uses it only for action selection — rohanpaul_ai · 2026-08-23
- China's AI robot progress over one year — iamfakhrealam · 2026-08-23
- The All Spark Cluster: Upgrading from 16 to 36 DGX Sparks — Kurcide · 2026-08-23