Mars 2020, NASA’s next and yet-to-be-named Mars rover, will be the first mission to collect and prepare samples of the martian surface for return to Earth. This process is known as caching, and it is the crucial first step of a fully-born sample return campaign that could define the next two decades of robotic Mars exploration. Recently, the Mars 2020 engineering team proposed a new caching strategy that differs from previous concepts in some interesting ways.
JPL calls this adaptive caching, but I like to think of it more as the cache depot strategy. This means that after coring samples and placing them into hermetically-sealed tubes (the same process for any sort of caching), the rover will then deposit groups of samples on the ground throughout its drive. A future rover would retrieve some or all of these samples, place them in a rocket, and launch them into Mars orbit.