Task 1 required that we rotate the robot in place to locate a verbally-specified object (banana) in its camera FOV, navigate toward it using visual servoing to keep it centered, stop at a fixed distance, compute a top-down grasp pose from the object's bounding box and point cloud, grab the object, lift it to a stowed sleep pose, and navigate back to the start position.

Task 2 required that we extend Task 1 with a navigation-and-deposit sequence: after retrieving the object (banana), the robot rotates to locate a verbally-specified destination container (bowl), visually servos toward it, stops at distance, extends the arm from the stowed pose to a forward position, and opens the gripper to drop the payload in.

Task 2.5 required a generalization of Task 2 to a more interesting, collaborative endpoint. After locating and retrieving the banana as in the prior tasks, the robot locates the new requested target: a person. Finally, the robot navigates towards the person and performs a legible handoff.