Robot Ethics and Models of Robot Environment Interactions

The robotic system observable behavior results from the coordinated working of its bodily parts and their interactions with the environment. The contribution of the environment in shaping robotic behaviors can be hardly overestimated. The trajectory of a personal robot negotiating a living room surface can be affected significantly by changes of frictional coefficients only, such as those introduced by a Persian carpet or a glass of water spilled on the floor. The paths of twisted traced on a beach by an insect-like.

Robotic engineers are well aware of the practical and theoretical difficulties surrounding the epistemic problem of identifying environmental disturbing factors which jeopardize the normal working of robotic systems. A strategy of heuristic which is often applied to address this epistemic problem is to make concrete environments in which robotic systems will be immerses as similar as possible to ideal situations in which suitable behavioral regularities are believed to hold. This strategy of heuristic relies on the fact that any consistent (non-contradictory) set of statements T admits a model, in the usual sense of the word 'model' which is adopted in mathematical logic.

More specifically, if T is a consistent set of sentences, then there is an interpretation of T, relative to some domain of objects and relations, which makes true all the sentences of T. This strategy of heuristic can be rationally reconstructed as a process involving two main steps.
1. Idealized domains of objects and relations are introduced, such that the desired regularities concerning robotic behaviors are true when interpreted in those domains;

2. The concrete environments in which robots will be immersed are modified in order to make them as similar as possible to the idealized domains introduced in the previous step.

One extensively applies the above strategy in order to enforce ideal-world conditions which exclude “disturbing factors” while preserving task compliance conditions in industrial automation. Since human workers are a major source of dynamic changes and disturbing factors impinging on industrial robot behaviors, a robot “segregation” policy is generally pursued to achieve quasi-static and more easily predictable robot environments: factory workers and robots are often confined to different workspaces, and their mutual interactions are severely limited or altogether excluded.