The Robot in the Deep End
1The robotics club at Fernhill Middle School is building a small underwater robot to explore the deep end of the town pool. The robot carries a camera, and a cable called a tether, 25 m long, connects it to a laptop at the poolside. The tether carries the camera's picture to the laptop as an electric signal. Club members Kenji Oduya and Paz Albright noticed a problem on their first test dive: whenever the pool's pump was running, white specks like snow flickered across the picture. Their adviser, Ms. Ingrid Tallis, explained that the pump's motor adds random changes in voltage, called noise, to signals in nearby cables.
2A camera picture is made of thousands of tiny squares called pixels. The robot's camera gives each pixel a brightness level from 0 (black) to 15 (white). The club's first version sends each brightness level as an analog signal: one voltage, 0.3 V for each level. So level 4 is sent as 1.2 V and level 15 as 4.5 V. The laptop divides the voltage it receives by 0.3 and rounds to the nearest whole level.
3The second version sends each level as a digital signal: a group of four pulses, each either on (5 V) or off (0 V). The four pulses stand for 8, 4, 2 and 1, in that order, and the laptop adds the values of the pulses that are on. For example, off, on, off, on stands for 4 + 1 = 5. The laptop reads any pulse above 2.5 V as on and any pulse at 2.5 V or less as off.
4To compare the two versions, the club placed a test card with a known pattern of gray squares on the pool floor and sent the robot down. For each test, the laptop checked a 1,000-pixel patch of the picture. A pixel counted as wrong when the laptop showed a level different from the one on the card. The club tested both versions with the pump off and on its low and high settings, using the same tether, card and lighting. Ms. Tallis measured the largest noise spike on the tether for each setting.
5Kenji then wondered whether a longer tether would let the robot reach the far end of the pool. The club added a 25 m extension, for 50 m in all, and repeated the high-pump test. The longer cable ran close to the pump room and picked up stronger spikes. Paz recorded every result in Table 1.