As a primary focus in deep-space exploration, cis-lunar space is expected to witness a surge in space activities, inevitably increasing collision risks. This study addresses this challenge by proposing a novel optimization algorithm for collision avoidance maneuvers using low-thrust propulsion under the circular restricted three-body problem (CRTBP). An efficient collision avoidance optimization algorithm is developed, leveraging a collision probability model to quantify uncertainties in the three-body environment. Using a homotopy approach, the fuel-optimal solution is derived. Simulations are conducted to analyze the effectiveness and feasibility of different maneuver initiation times. The results demonstrate that the proposed algorithm achieves a reduction in fuel consumption while maintaining collision avoidance effectiveness, providing an optimal solution with practical engineering significance for three-body collision avoidance maneuvers. © 2025 Elsevier B.V., All rights reserved.