TubeDAgger: Reducing the Number of Expert Interventions with Stochastic Reach-Tubes
This work addresses the efficiency of training novice policies in robotics or autonomous systems by reducing reliance on human experts, though it is incremental as it builds on the DAgger framework.
The paper tackles the problem of reducing expert interventions in interactive imitation learning by introducing stochastic reachtubes to estimate when expert control is needed, resulting in fewer interventions compared to doubt classification methods without requiring per-environment threshold tuning.
Interactive Imitation Learning deals with training a novice policy from expert demonstrations in an online fashion. The established DAgger algorithm trains a robust novice policy by alternating between interacting with the environment and retraining of the network. Many variants thereof exist, that differ in the method of discerning whether to allow the novice to act or return control to the expert. We propose the use of stochastic reachtubes - common in verification of dynamical systems - as a novel method for estimating the necessity of expert intervention. Our approach does not require fine-tuning of decision thresholds per environment and effectively reduces the number of expert interventions, especially when compared with related approaches that make use of a doubt classification model.