Robot-assisted, multi-axis additive manufacturing enhances fabrication flexibility compared to conventional 3D printing by enabling non-planar material deposition. This capability is particularly relevant for wire arc additive manufacturing (WAAM), which offers cost-effective construction of large metallic components. Non-planar fabrication introduces a new appraoch to addressing challenges such as residual stresses and distortions in WAAM, as the fabrication sequence has a high impact on their accumulation. To analyse these effects, space-time optimization [1-2] provides a computational framework that encodes fabrication sequences with a pseudo-time field. The iso-surfaces of this scalar field define the (curved) layers. While promising, the layers generated by this approach often exhibit irregular geometries, making them difficult to fabricate in practice. To improve manufacturability, we introduce a novel approach for regularizing the geometry of individual layers by controlling both thickness variation and curvature. Maintaining an (almost) constant layer thickness minimizes frequent and abrupt changes in processing parameters, such as move speed. Additionally, curvature control prevents the formation of concave layers that may be inaccessible (i.e., risking potential collision with the print head). Numerically, layer thickness is approximated by the gradient of the pseudo-time field, while the curvature is computed by the Hessian matrix. As shown in Fig. 1, incorporating geometric regularization into fabrication sequence optimization improves layer regularity while reducing distortion compared to planar layers.