<p>Constrained multi-objective optimization problems are ubiquitous in real-world applications. Efficiently solving these problems necessitates a delicate equilibrium between feasible and infeasible solutions. In recent years, numerous constraint handling techniques have been proposed by scholars, primarily aimed at enhancing algorithmic solution performance. However, within complex feasible domains, an excessive reliance on methodologies that are solely geared towards addressing feasible or infeasible solutions has been shown to inhibit the efficiency of the algorithmic process of exploration. Seeking to leverage both solution types potential, this paper presents CMATA—an innovative tri-population constrained multi-objective evolutionary algorithm. The first two population evolve driven by feasibility; Specifically, the first population evolves based on the original problem, while the auxiliary second population optimizes M objective functions for the original problem, with constraint violations considered as an additional (M + 1) objective. A third population was evolved using an improved constraint relaxation technique. Throughout the evolutionary process, these three populations identify the constrained search area and collaborate to store viable options from the search process in an external archive. The efficacy of CMATA was tested against five advanced algorithms on 37 benchmark problems and a real-world problems. The results of these empirical tests confirm CMATA efficacy and competitive advantage in solving constrained multi-objective optimization problems.</p>