Optimizing the traffic flow in a city is a challenging problem, especially in a future traffic system of self-driving cars. This is due to the interactions between the individual traffic agents (vehicles) who compete for the use of the common infrastructure (streets) given traffic dynamics such as stop-andgo effects, changing lanes, and other. The goal of this paper is to provide a solution to the above problem that works in a fully decentralized and participatory way, i.e. autonomous agents collaborate without a centralized data collector and arbitrator. Such a solution should be scalable, privacy-preserving, and flexible with respect to the degree of autonomy of agents. A self-adaptive framework to support this research is introduced: TRAPP – Traffic Reconfigurations via Adaptive Participatory Planning. The framework relies on a microscopic traffic simulator, SUMO, for simulating urban mobility scenarios, and on a decentralized multi-agent planning system, EPOS, for decentralized combinatorial optimization, applied here in traffic flows. A data-driven inter-operation of the two tools in our framework allows high modularity and customization for experimenting with different scenarios, optimization objectives and agents’ behavior and as such providing new perspectives for resilient infrastructures for self-driving cars.