What you basically doing is to get all hardware set up ready for your robot ...you just running a bunch of ros nodes /pacakges (Gmapping,hector ,amcl,move_base etc) on your brain of your robot(master) ..make sure you get a good dc motor with attached encoder .They are alot you can get online ,also the packages your running doesn't determine the hardware to get. You just connecting both publisher and subscriber node which has a topic name to integrate ros with this hardware via a communication protocol. As for hector node you don't need a means of odom to publish a map..you can use a lidar/depth camera to generate a map since the tf tree goes from map>base_link>laser_frame .But that of gmapping you need odom data to get it started ...in all view ,get a great hardware that suits ros implementation.