A heterogeneous Social Force Model for Personal Mobility Vehicles on futuristic sidewalks

Electric scooters are becoming popular in public spaces, and autonomous robots will join soon. However, integrating these Personal Mobility Vehicles (PMV) without proper provisions challenges the safety and comfort of all users. While Social Force Model (SFM) commonly replicates pedestrians' movements, directly applying it to PMVs is challenging and inaccurate. We propose a heterogeneous SFM considering the dynamic personal spaces of various agents on futuristic sidewalks, addressing the impracticalities of SFM. Additionally, subjective safety estimation relaxes the constant desired-velocity assumption, and the influence weight reduces the complexity by omitting pairwise calibration. Experiments calibrate the model for e-scooters, validating it in realistic scenarios with multiple e-scooters passing through pedestrians. The proposed model has higher accuracy than previous models regarding behavioral naturalness metrics. In addition, the models' performance in replicating experimental observations is analyzed. This research contributes to safer and more efficient transportation with PMVs, particularly e-scooters, and provides a novel approach to modeling multi-type agents on heterogeneous sidewalks.