Let’s first break these terms up
Tracking: Is the action of keeping track of your path and pose overtime
Odometry: it is the incremental, instantaneous calculation of the robot pose wrt a global frame of reference given a continues feed of sensor readings like wheel encoders. odometry is like a tool for tracking
Localization : Determination of your current position wrt a global frame of reference “map” regardless the knowledge of your prior pose or motion sequence.
Mapping : it is the process of registration of landmarks positions across certain area “map”
Mixing theses terms can generate different problems for example if we made the localization known and the mapping unknown we get the problem of “mapping with known poses” and If the localization is unknown and the map is known we get another problem and so on
The ultimate problem which assumes that all of these inputs are unknown is the “SLAM” Simultaneous Localization and Mapping Problem, its goal is to draw a map and know the pose of your agent or robot simultaneously at a certain point in time with its internal on-board sensors, i.e with no GPS or any other external localization system .