Community Discussion · Policy

AI in 3D Manufacturing: What RL Researchers Should Focus On

Dao Shi Shuo DuiJul 252026/07/25 63 views

In Q2 2024, global funding for 3D printing startups dropped 12% quarter-over-quarter, but the AI 3D generation sector bucked the trend with 34% growth. The Hi3D closed-door meeting was held at this juncture, where Ren Lifeng publicly discussed the roadmap from digital generation to physical manufacturing for the first time. As someone researching reinforcement learning, I see my field being pulled into a more concrete engineering problem.

Policy

1 replies

Ctrl + Enter to reply

Professional BuzzkillJul 28(edited)

[quote="zhong_yiming, post:1, topic:1619"]

In Q2 2024, global funding for 3D printing startups dropped 12% quarter-over-quarter, but the AI 3D generation track grew 34% against the trend. Hi3D held a closed-door meeting at this juncture, where Ren Lifeng publicly discussed the roadmap from digital generation to physical manufacturing for the first time. As someone researching reinforcement learning, I see my field being drawn into a more specific engineering problem.

Short-term, AI 3D is stuck in the crack between "generation" and "manufacturing." Generating 3D models using diffusion models or NeRF is already mature; you can get a mesh from a single image. But Hi3D disc…

[/quote]

Hard-learning offline reinforcement learning from print logs—the inconsistent data quality is the biggest pitfall. Moreover, the sim-to-real gap is even larger than for robotic arms. One print takes hours, making sample efficiency unsolvable. This direction is overheating again.