The effects of implementing human valence into the behavioral model of a fully autonomous vehicle

Objective: To test in two experiments whether people would be more willing to use a Level-5 Fully Autonomous Vehicle (FAV) if the FAV's driving model is enhanced with human valence. Background: Although it is believed that FAVs will eventually replace manual vehicles, people at present worry about the safety of FAVs, sense discomfort about FAVs, and it is doubtful whether they would be willing to use FAVs. Human valence is the degree to which someone is drawn towards or repulsed away from an object. However, the literature lacks empirical support on whether enhancing FAVs with valence could help ease concerns about safety, discomfort, and willingness to use FAVs. Method: In Experiment I, conducted to address this void, participants manually drove in a driving simulator a simulated route at either 50 or 100 km/h and overtook road objects that traveled at either 0 or 50 km/h. Three proxy measures were recorded to infer human valence toward the overtaken objects: time-to-collision, lateral clearance, and perceived risk. The inferred valence was then implemented into Experiment II's driving model of a FAV that different participants drove. Results: Participants were most willing to re-use a Safe driving style. However, the enhanced FAV's willingness was not significantly different from the Safe driving style. Conclusion: The first empirical evidence about enhancing the driving model of FAVs with human valence motivates further research. Application: FAV manufacturers are advised to include human valence in the driving model of FAVs to increase people's willingness to use them.