Automating Transfer of Robot Task Plans Using Functorial Data Migrations

This paper introduces a novel approach to ontology-based robot plan transfer by leveraging functorial data migrations, a structured mapping method derived from category theory. Functors provide structured maps between planning domain ontologies which enables the transfer of task plans without the need for replanning. Unlike methods tailored to specific plans, our framework applies universally within the source domain once a structured map is defined. We demonstrate this approach by transferring a task plan from the canonical Blocksworld domain to one compatible with the AI2-THOR Kitchen environment. Additionally, we discuss practical limitations, propose benchmarks for evaluating symbolic plan transfer methods, and outline future directions for scaling this approach. Note to Practitioners—This research was motivated by the challenge of reusing successful robot task plans in different domains without the need to redesign or recalibrate the plans from scratch. In industries where automation is prevalent, such as manufacturing or logistics, the ability to efficiently transfer a task plan from one setting to another can significantly enhance productivity and reduce operational costs. Our solution utilizes functorial data migrations, a concept from category theory, to enable the structured transfer of task plans between distinct domains. This approach ensures that transferred plans remain valid and applicable. For example, a plan developed to transport items in a warehouse can be adapted to transport medical supplies in a hospital, which leverages the underlying structure of the tasks without altering their essence. While our approach generalizes task plan transfer across domains, it relies on precise domain ontology maps, which can be complex. This method may not capture all nuanced transfers between highly-specialized domains, particularly where domain-specific knowledge is comprehensive. Practitioners should be aware that some manual tuning might still be necessary to fully optimize the plans for specific operational contexts.