Build on Priors: Vision--Language--Guided Neuro-Symbolic Imitation Learning for Data-Efficient Real-World Robot Manipulation

1. 1Construct symbolic planning domains autonomously
2. 2Utilize vision-language guidance for neuro-symbolic imitation learning
3. 3Leverage handful of demonstrations for long-horizon manipulation tasks
4. 4Integrate learned models with real-world robot manipulation systems

Robotics engineers and AI researchers on a team can benefit from this framework as it enables robots to learn complex tasks with minimal demonstrations, improving overall efficiency and scalability.

💡 Autonomously constructing symbolic planning domains enables scalable and data-efficient real-world robot manipulation

🤖 Data-efficient robot manipulation with neuro-symbolic learning!