Achieving full autonomy for a planetary explorer is the main requirement in rendering feasible missions when the communication time with the ground station does not allow realtime operation and monitoring. The design process involved in building the MarsWorks rover illustrates the challenges to be addressed in a typical surface exploration mission. This paper presents the main stages necessary to achieve rover autonomy in Mars-analogue environments. The focus is on two key areas: rough terrain navigation and autonomous manipulation with a six degree-of-freedom robotic arm. The first topic covers fundamental data fusion and Kalman filtering methods that estimate the current pose, as well as displacements from the starting position by means of visual-inertial odometry. An approach to guidance and control is then presented from the perspective of the dynamic window technique. Subsequently, autonomous grasping with increasing levels of automation is presented: from the low-level proportional–integral–derivative (PID) control to inverse kinematics, motion planning, computer vision, and automatic target recognition. Finally, onboard data handling, fusion of the sensor data used for scientific sample analysis, and communication with the ground station are briefly discussed. Each section presents future ambitions and possible ways of optimising individual subsystems of the MarsWorks rover.