Using Abstraction for Interpretable Robot Programs in Stochastic Domains
This addresses the challenge of comprehensibility in robot programming for researchers and practitioners in robotics and AI, though it appears incremental as it builds on existing Golog extensions.
The paper tackled the problem of interpretability in robot programs for stochastic domains by using abstraction to create high-level nonstochastic models, resulting in programs that are easier to understand, often eliminate belief operators or loops, and produce shorter action traces.
A robot's actions are inherently stochastic, as its sensors are noisy and its actions do not always have the intended effects. For this reason, the agent language Golog has been extended to models with degrees of belief and stochastic actions. While this allows more precise robot models, the resulting programs are much harder to comprehend, because they need to deal with the noise, e.g., by looping until some desired state has been reached with certainty, and because the resulting action traces consist of a large number of actions cluttered with sensor noise. To alleviate these issues, we propose to use abstraction. We define a high-level and nonstochastic model of the robot and then map the high-level model into the lower-level stochastic model. The resulting programs are much easier to understand, often do not require belief operators or loops, and produce much shorter action traces.