Image-guided robot-assisted minimally invasive surgery is an important medicine procedure used for biopsy or local target therapy. In order to reach the target region not accessible using traditional techniques, long and thin flexible needles are inserted into the soft tissue which has large deformation and nonlinear characteristics. However, the detection results and therapeutic effect are directly influenced by the targeting accuracy of needle steering. For this reason, the needle-tissue interactive mechanism, path planning, and steering control are investigated in this review by searching literatures in the last 10 years, which results in a comprehensive overview of the existing techniques with the main accomplishments, limitations, and recommendations. Through comprehensive analyses, surgical simulation for insertion into multi-layer inhomogeneous tissue is verified as a primary and propositional aspect to be explored, which accurately predicts the nonlinear needle deflection and tissue deformation. Investigation of the path planning of flexible needles is recommended to an anatomical or a deformable environment which has characteristics of the tissue deformation. Nonholonomic modeling combined with duty-cycled spinning for needle steering, which tracks the tip position in real time and compensates for the deviation error, is recommended as a future research focus in the steering control in anatomical and deformable environments. Graphical abstract a Insertion force when the needle is inserted into soft tissue. b Needle deflection model when the needle is inserted into soft tissue [68]. c Path planning in anatomical environments [92]. d Duty-cycled spinning incorporated in nonholonomic needle steering [64].