Neural Task Programming: Learning to Generalize Across Hierarchical Tasks

In this work, we propose a novel robot learning framework called Neural Task Programming (NTP), which bridges the idea of few-shot learning from demonstration and neural program induction. NTP takes as input a task specification (e. g. , video demonstration of a task) and recursively decomposes it into finer sub-task specifications. These specifications are fed to a hierarchical neural program, where bottom-level programs are callable subroutines that interact with the environment. We validate our method in three robot manipulation tasks. NTP achieves strong generalization across sequential tasks that exhibit hierarchal and compositional structures. The experimental results show that NTP learns to generalize well towards unseen tasks with increasing lengths, variable topologies, and changing objectives. stanfordvl.github.io/ntp/.

Authors

• Danfei Xu
• Suraj Nair 0003
• Yuke Zhu
• Julian Gao
• Animesh Garg
• Li Fei-Fei 0001
• Silvio Savarese

Keywords

• Task analysis
• Programming
• Robots
• Sorting
• Semantics
• Topology
• Data models
• Hierarchical Task
• Specific Tasks
• Manipulation Tasks
• Neural Induction
• Inverse Reinforcement Learning
• Robot Learning
• Time Series
• Semantic
• Real-world Tasks
• Object Task
• Long Short Memory
• Prolonged Interaction
• Sorts Of Objects
• Block Stacking