I did try several solutions to including empty groups in my Kanban view.

First I did copy more or less the OSV solution from the Odoo Project module. But that didn't work:

def _read_group_stage_ids(self, cr, uid, ids, domain, read_group_order=None, access_righ        ts_uid=None, context=None):    
    if context is None:        context = {}    
    stage_obj = self.pool.get('project.task.type')    
    order = stage_obj._order    
    access_rights_uid = access_rights_uid or uid    
    if read_group_order == 'state_def desc':        
        order = '%s desc' % order
    if 'default_project_id' in context:
         search_domain = ['|', ('project_ids', '=', context['default_project_id']), ('id            ', 'in', ids)]    
    else: 
         search_domain = [('id', 'in', ids)]    
    stage_ids = stage_obj._search(cr, uid, search_domain, order=order, access_rights_uid             =access_rights_uid, context=context)    
    result = stage_obj.name_get(cr, access_rights_uid, stage_ids, context=context)        # restore order of the search    
    result.sort(lambda x, y: cmp(stage_ids.index(x[0]), stage_ids.index(y[0])))    
    fold = {}    
       for stage in stage_obj.browse(cr, access_rights_uid, stage_ids, context=context):       fold[stage.id] = stage.fold or False    
    return result, fold
_group_by_full = {    'stage_id': _read_group_stage_ids,    }

But then I found a simpler solution according the tutorial from Ludwik Trammer.

\http://ludwiktrammer.github.io/odoo/odoo-grouping-kanban-view-empty.html

@api.model
def _read_group_stage_ids(self, present_ids, domain, **kwargs):    
    # project_id = self._resolve_project_id_from_context()    
    result = self.env['project.task.type'].search([]).name_get()    
    return result, None

_group_by_full = {    'stage_id': _read_group_stage_ids,    }