Specification-Guided Data Aggregation for Semantically Aware Imitation Learning
This work addresses the challenge of enhancing imitation learning for autonomous driving by leveraging formal specifications, though it appears incremental as it builds on existing environment sampling techniques.
The paper tackles the problem of improving imitation-learned models in safety-critical scenarios like autonomous driving by using specification-guided sampling to aggregate expert data in semantically similar environments, resulting in more accurate models than other sampling methods in CARLA simulator experiments.
Advancements in simulation and formal methods-guided environment sampling have enabled the rigorous evaluation of machine learning models in a number of safety-critical scenarios, such as autonomous driving. Application of these environment sampling techniques towards improving the learned models themselves has yet to be fully exploited. In this work, we introduce a novel method for improving imitation-learned models in a semantically aware fashion by leveraging specification-guided sampling techniques as a means of aggregating expert data in new environments. Specifically, we create a set of formal specifications as a means of partitioning the space of possible environments into semantically similar regions, and identify elements of this partition where our learned imitation behaves most differently from the expert. We then aggregate expert data on environments in these identified regions, leading to more accurate imitation of the expert's behavior semantics. We instantiate our approach in a series of experiments in the CARLA driving simulator, and demonstrate that our approach leads to models that are more accurate than those learned with other environment sampling methods.