Indication of interaction plans based on model predictive interaction control: Cooperation between AMRs and pedestrians using eHMI

Abstract

Interactions between autonomous mobile robots (AMRs) and pedestrians can be difficult due to the lack of communication tools: languages, gestures, and eye contact. An AMR’s dynamic behavior and the use of external human-machine interfaces (eHMIs) are expected to be alternatives for these tools. However, most of the previous studies do not consider pedestrians’ intentions, thus the plan of interaction may be disagreeable for pedestrians. This study focuses on the process of making an agreement with pedestrians. By estimating pedestrians’ intentions with model predictive control (MPC), optimal plans which are acceptable for pedestrians are calculated. The plan is then displayed by using an eHMI in order to communicate and reach an agreement on the future interaction plan with pedestrians. Results of a human-in-the-loop simulation (HiLS) show that the proposed eHMI can increase smooth interactions. This enables AMRs and pedestrians to pass by each other quickly and resolves the uneasiness pedestrians feel when interacting with AMRs.