Adaptive Preferences: Pivoting Through User Complaints

In the evolution of software systems, especially in domains like autonomous vehicles, dynamic user preferences are critical yet challenging to accommodate. Existing methods often misrepresent these preferences, either by overlooking their dynamism or overburdening users as humans often find it challenging to express their objectives mathematically. Addressing this, we introduce a novel framework interpreting dynamic preferences as inherent uncertainty, anchored on a “human-on-the-loop” mechanism enabling users to give feedback when dissatisfied with system behaviors. Leveraging a designed fitness function, our system employs a genetic algorithm to adapt preference values, aligning preferences with user expectations through feedback-driven adaptation. We validated its effectiveness with an autonomous driving prototype and a user study involving 20 participants. The findings highlight our framework’s capability to effectively merge algorithm-driven adjustments with user complaints, leading to improved participants’ subjective satisfaction in autonomous systems.