Sampling point clouds with N points

I am trying to sample a point cloud dataset with the same number of points.

The solution I thought was to convert the point clouds into mesh and use read_triangle_mesh(i).sample_points_uniformly(N_POINTS) however I get problems converting some clusters:

mesh = o3d.geometry.TriangleMesh.create_from_point_cloud_alpha_shape(pcd_i, alpha=0.01)
RuntimeError: QH6019 qhull input error: can not scale last coordinate. Input is cocircular
or cospherical. Use option 'Qz' to add a point at infinity.

While executing: | qhull d Qbb Qt
Options selected for Qhull 2018.0.1.r 2018/12/28:
run-id 1378539865 delaunay Qbbound-last Qtriangulate _pre-merge
_zero-centrum Pgood

Is there other method? Is it a bug?