Unverified50% confidenceFactExact time
Post-training methods require large amounts of expensive real robot data, and intensive post-training must be redone for every change of base model.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedExact unlearning achieves complete deletion by retraining the model after removing the target data, which is extremely costly for large models with billions of parameters, as a single complete training run may consume millions of dollars in computing resources.77% similarUnverifiedCompanies such as NVIDIA have in recent years heavily invested in simulation training and synthetic data, having robots practice millions of times in virtual environments to compensate for the lack of real-world data.76% similarUnverifiedTraining reinforcement learning directly on real robots is extremely costly, since every fall may damage hardware and sampling efficiency is low, so simulation environments (such as Isaac Sim, MuJoCo and Gazebo) provide low-cost, parallelizable, resettable training grounds.75% similarUnverifiedIn recent years, the trend in robotic foundation models has been to leverage large-scale pretraining to improve generalization, then adapt to specific tasks through fine-tuning.74% similarUnverified四足机器人的自我恢复能力来自强化学习的sim-to-real训练范式，通过在仿真环境中摔倒数百万次学习平衡策略72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130295API
curl https://kongchang.com/api/v1/knowledge/claims/1130295MCP
get_claim(id=1130295)