We recently introduced a new and innovative framework named “in vivo computation” by modeling the tumor targeting problem as a natural computation problem. Nanorobots play the role of computing agents are guided by the information of biological gradient field (BGF) induced by the emerging of tumor for the searching of tumor location which is the optimal solution for the in vivo computational problem. To overcome the in vivo constraints encountered in previous research, which primarily concentrated on tumor detection, several computational strategies have been suggested for achieving tumor targeting. This work concentrates on the exploration of tumor boundary with nanorobots, which is a novel and valuable research point. To overcome this challenge, we resort to the spontaneous motion of nanorobots in liquid environment, where the local hydrodynamic flows are used to actuate the nanorobots to keep a balance to the effect of BGF. In order to showcase the efficacy of the proposed methods, we conduct in silico experiments across three BGF landscapes that vary in terms of their optimization complexity.