<p>The octopus has evolved a stereotyped bending propagation mechanism to maneuver its highly flexible arms underwater, using suckers for sensing and capturing prey. Inspired by this biological strategy, we propose an activation propagation-based model that generates motion through phase differences in the activation of dorsal and ventral muscles. This model requires only three input parameters to achieve reaching behavior, without the need for full-shape control. We applied this model to a six-segment soft continuum robot and conducted experiments that confirmed the low-resistance characteristics of underwater bending propagation locomotion. To enable autonomous interaction, we integrated suckers with attachment-sensing capabilities and developed an “approach-sense-grasp” strategy, allowing the robot to retrieve floating objects both in open water and within confined environments. Our findings shed light on the control principles behind octopus arm locomotion and demonstrate their applicability in simplifying control of soft robotic systems for efficient, adaptive grasping in underwater environment.</p>