Robot learning still splits into Sim2Real for locomotion and BC for manipulation
nwilliams030 · x · 2026-07-21
Monday’s robot-learning thesis: Sim2Real and BC still define the field
The post argues that, since deep learning, robot learning has seen two foundational shifts: Sim2Real for locomotion and Behavior Cloning (BC) for manipulation.
The attached chart makes a second point: post-training methods such as RL or DAgger can keep improving performance after the initial training curve starts to flatten, but startups may hit the VC patience limit before they fully cross the ROI line. In other words, the technical flywheel may be real, but the business window can close first.
Related event: Two Core Paradigms in Robot Learning: Sim2Real and BC(2 posts)→
More from Embodied
- Johns Hopkins Launches Full-Stack Hands-on Robot Learning Class with SO-101 Arm Kits — _krishna_murthy · 2026-09-11
- SyncWorld: In-Context Robot World Model Simulates Unseen Views and Embodiments Zero-Shot — ChongZzZhang · 2026-09-11
- A 3D Pose Dataset for Dogs Released — ducha_aiki · 2026-09-11
- Swaayatt demos autonomous driving at 52 km/h on mountain roads, self-recovers after skid — sanjeevs_iitr · 2026-09-11
- AUAR's MicroFactory brings a deployable robotic wood-panel factory to the construction site — lukas_m_ziegler · 2026-09-11
- Musk: Cybercab certified at 165 Wh/mi, the most efficient production EV ever — elonmusk · 2026-09-11