This paper introduces an Evolution-Guided Bayesian Optimization (EGBO) algorithm for constrained multi-objective optimization in self-driving labs. The algorithm integrates selection pressure with a q-Noisy Expected Hypervolume Improvement (qNEHVI) optimizer to efficiently solve for the Pareto Front (PF) and achieve better coverage of the PF while limiting sampling in the infeasible space. The algorithm is developed together with a custom self-driving lab for seed-mediated silver nanoparticle synthesis, targeting three objectives: (1) optical properties, (2) fast reaction, and (3) minimal seed usage, alongside complex constraints. The study demonstrates that EGBO outperforms state-of-the-art qNEHVI in performance, showing significant hypervolume improvement across various synthetic multi-objective problems. EGBO also demonstrates good coverage of the PF and a better ability to propose feasible solutions. The paper discusses the challenges of constrained multi-objective optimization (cMOOPs) in materials science and chemistry, where multiple objectives and constraints must be satisfied. It highlights the need for efficient sampling in high-throughput experimentation platforms due to limited evaluation budgets. The paper also addresses the difficulty of handling constraints that make some trade-off solutions potentially impossible to evaluate, and the need for an ideal optimization algorithm that efficiently exploits towards the PF, uniformly explores the PF, and avoids infeasible regions near the PF. The paper presents a comparison between qNEHVI-BO and EGBO, showing that EGBO outperforms qNEHVI-BO in terms of optimization efficiency and constraint handling. EGBO is able to bypass infeasible regions near the PF, even in high-dimensional decision spaces with very limited feasible space. The paper also discusses the performance of EGBO on synthetic problems, showing that it outperforms qNEHVI-BO in terms of hypervolume, non-uniformity, and sampling wastage. The paper also discusses the handling of input and output constraints, showing that EGBO is effective in both cases. The paper concludes that EGBO is a general framework for efficiently solving constrained multi-objective problems in high-throughput experimentation platforms. It is particularly useful for industry-relevant problems where experiment time and chemical reactants are limited. The paper also discusses the importance of properly integrating the repair operator into the overall optimization workflow when handling input constraints. The paper proposes a strategy to pre-repair these constraints, which is shown to be more effective than post-repair. The paper also discusses the potential for future work, including extending EGBO to handle asynchronous batching.This paper introduces an Evolution-Guided Bayesian Optimization (EGBO) algorithm for constrained multi-objective optimization in self-driving labs. The algorithm integrates selection pressure with a q-Noisy Expected Hypervolume Improvement (qNEHVI) optimizer to efficiently solve for the Pareto Front (PF) and achieve better coverage of the PF while limiting sampling in the infeasible space. The algorithm is developed together with a custom self-driving lab for seed-mediated silver nanoparticle synthesis, targeting three objectives: (1) optical properties, (2) fast reaction, and (3) minimal seed usage, alongside complex constraints. The study demonstrates that EGBO outperforms state-of-the-art qNEHVI in performance, showing significant hypervolume improvement across various synthetic multi-objective problems. EGBO also demonstrates good coverage of the PF and a better ability to propose feasible solutions. The paper discusses the challenges of constrained multi-objective optimization (cMOOPs) in materials science and chemistry, where multiple objectives and constraints must be satisfied. It highlights the need for efficient sampling in high-throughput experimentation platforms due to limited evaluation budgets. The paper also addresses the difficulty of handling constraints that make some trade-off solutions potentially impossible to evaluate, and the need for an ideal optimization algorithm that efficiently exploits towards the PF, uniformly explores the PF, and avoids infeasible regions near the PF. The paper presents a comparison between qNEHVI-BO and EGBO, showing that EGBO outperforms qNEHVI-BO in terms of optimization efficiency and constraint handling. EGBO is able to bypass infeasible regions near the PF, even in high-dimensional decision spaces with very limited feasible space. The paper also discusses the performance of EGBO on synthetic problems, showing that it outperforms qNEHVI-BO in terms of hypervolume, non-uniformity, and sampling wastage. The paper also discusses the handling of input and output constraints, showing that EGBO is effective in both cases. The paper concludes that EGBO is a general framework for efficiently solving constrained multi-objective problems in high-throughput experimentation platforms. It is particularly useful for industry-relevant problems where experiment time and chemical reactants are limited. The paper also discusses the importance of properly integrating the repair operator into the overall optimization workflow when handling input constraints. The paper proposes a strategy to pre-repair these constraints, which is shown to be more effective than post-repair. The paper also discusses the potential for future work, including extending EGBO to handle asynchronous batching.