I believe I read in a book once a description of a kind of interplanetary ferry that would orbit the sun and regularly pass different planets like Mars and Earth. The idea of the ferry is to have a place to stay for astronauts on their long journey apart from being strapped into the confines of a single space ship. I find it hard, however, to find anything on the web on this idea. Would it be possible for such a ferry to frequent more than two planets with a single orbit?
Mars cyclers have already been mentioned in the comments, with the Aldrin cycler being one example of them. The castle parts of them could be seen as ferries, and the visiting taxis as means of embarking and disembarking during visits of planets along their routes.
And three or more bodies cyclers are also possible, for example the so-called resonant cyclers of which Laplace resonant cyclers in the Jupiter system might be the best example and are made possible due to the Laplace orbital resonance between Io, Europa and Ganymede (see e.g. Laplace-resonant triple-cyclers for missions to Jupiter, A. E. Lynam et al.). These can be either gravity assist or visit cyclers.
There could also be visit only non-resonant cyclers with highly elliptical heliocentric orbits that stretch orbital altitude of several planet but never intersect their orbits (slight inclanation of the cycler w.r.t. orbital plane of involved planets), but those would be less useful in terms of delta-v savings for the taxis to and fro planets, tho they could still provide significant mass savings with their reusable castle parts, depending on utility they could provide (see e.g. this thread for some suggested uses).
Of course, there might be an infinite number of cyclers and other periodic or quasi-periodic orbits between multiple celestial bodies. Problem with finding those is the computational intensity mostly limiting analysis to non-exhaustive R3BP (restrictive three-body problem) and taking shortcuts such as restricting search to best possible candidates, natural equilibrium between celestial bodies, and using precomputed invariant manifolds in (simplified) R3BP space, then stitching these together to design more complex trajectories involving multiple targets. Adding more bodies to such analysis greatly intensifies search for additional possible resonant cyclers (or other low energy trajectories, often the lot of them known under a moniker of Interplanetary Transport Network) within our Solar system, and is still one of the areas where human intuition, for lack of a better word, often trumps all the supercomputers that we task with solving such problems.
Throw at this problem also the possibility of using propulsion methods that don't consume any reaction mass, such as momentum exchange tethers, solar or magnetic sails, beam powered propulsion, ramjets and similar, or settle for shorter term use or that your ferry system doesn't have to be self-sufficient (dependent on supply of consumables and would require maintenance), and the number of possibilities becomes infinite.