Abstract: To address the issue of interference with ultrasound imaging stability caused by null-space collision avoidance reactions in the 7-DOF ultrasound robot-assisted puncture task, we propose a collision avoidance and image stabilization control method. Firstly, we establish a collision force and collision location estimation model under multi-point contact conditions, and propose a new collision threshold judgment method integrating external force and external torque to distinguish between collision situations that could and could not be avoided by null-space motion. Secondly, we design a dual-scale visual servo controller, incorporating the stability evaluation of the ultrasound image on the temporal scale and the displacement of feature blocks on the spatial scale into the robot's control loop. This controller is capable of dynamically adjusting the probe position to stabilize the ultrasound image. Finally, we conduct collision avoidance and image stabilization experiments on a 7-DOF ultrasound robot-assisted puncture platform. The experimental results show that the proposed method has an external force estimation error range of −4.33 N, 5.2 N during robot motion, and can effectively determine the collision situation to prevent the external force on the robot from rising to a large value, thereby ensuring the safety of physical human-robot interaction. Compared with the force/position hybrid control method, the proposed method achieves a 35.6% increase in the average value and a 48.5% increase in the maximum value of image stability evaluation during null-space collision avoidance. Moreover, it is capable of recovering the ultrasound images that have deviated from the desired state. These results validate the effectiveness of the proposed method.