A robot lives in many frames at once
A coordinate frame is just a fixed reference: an origin point plus a set of directions that define what "up," "forward," and "sideways" mean. A single robot typically works with several frames simultaneously. There's a world frame that describes the room or environment, a base frame attached to the robot's body, a camera frame attached to wherever its vision sensor sits, and an end-effector frame attached to its gripper or tool tip.
Each frame is convenient for a different purpose. A camera reports what it sees in its own camera frame, but a robot arm's controller needs to know where that object is relative to its base frame in order to plan a reach. Neither frame is wrong, they simply describe the same physical world from different vantage points.