清华AIR 首届博士李健雄创立本溯智能：五源资本领投

The team, including Tsinghua's Prof. Zhan Xianyuan, aims to reduce deployment costs significantly by enabling rapid adaptation to new environments.

Key Points

• BASAL's model focuses on action-native In-Context learning, diverging from traditional language-based approaches.
• The team includes 7 researchers who turned down high-paying offers totaling nearly 40 million yuan.
• BASAL's X-VLA model achieved top results in third-party evaluations with under 1000 hours of pre-training.
• The company aims to reduce the costs of deploying robots in varying environments significantly.
• BASAL is one of the first teams in China to explore rapid adaptation in embodied learning.