To address multi objective conflicts and insufficient decision space diversity in UAV path planning within complex three dimensional environments, this paper proposes an improved multimodal multi objective differential evolution algorithm, termed MMODE_HDMS. The algorithm formulates fuel consumption and comprehensive flight risk as the two competing optimization objectives and adopts a spherical vector coordinate representation for path encoding. Built upon the MMODE_CSCD framework, it incorporates a dynamic mutation probability strategy, a hybrid mutation operator, and a Hausdorff distance based environmental selection mechanism to balance exploration and exploitation, reinforce local search capability, and preserve the morphological diversity of candidate paths. Extensive experiments are conducted on four types of 3D simulation terrains, constructed by configuring distinct obstacle densities in both symmetric and asymmetric scenarios. The experimental results demonstrate that MMODE_HDMS reduces the IGD value by 20.0% and 39.4% under symmetric terrains, and by 5.9% and 10.7% under asymmetric terrains, compared with MMODE_CSCD. Furthermore, the proposed algorithm exhibits competitive performance in terms of distribution and diversity and can generate a set of equivalent optimal trajectories that achieve similar objective values yet differ significantly in geometric configuration. These findings substantiate the superiority and practical utility of MMODE_HDMS in offering diversified and trade off alternatives in real world reconnaissance missions.