Navigation¶
Global positioning system (GPS)¶
Navigation underwater is particularly challenging compared to surface and air based systems because there is no GPS. GPS requires the reception of satellite broadcasts, and so it is not possible to use this system underwater beneath a foot or so of water depth. This has been a topic of active research for a long time. It will probably remain so for some time to come, despite a few DARPA challenges designed to address the issue with ocean scale underwater positioning system using acoustics.
While it is hard to imagine a world without GPS at this point, in the 1980s access to the network was restricted, and it was only in 2000 that full accuracy of the positioning network was allowed for use by the civilians. There are still a few stipulations on use required in all GPS enabled devices to prevent its use in ballistic missile systems, and a standing contingency to revoke its use in wartime. This military history of the positioning system provides some hint to the strategic importance of GPS, and wartime contingencies continue to motivate GPS-free navigation for defense systems even when the system is readily available.
Dead Reckoning¶
Without the option of GPS, it is important to use other navigation techniques. A time tested method is dead-reckoning , used by sailors once they discovered an accurate compass. Dead-reckoning requires the measurement of the ships speed, which can be done by dropping a log off the bow and waiting for it to pass the stern. With a heading, it is possible to anticipate where the ship will be on a map without landmarks or celestial indicators, which can be very important in the open ocean with overcast skies. These basic instruments (compass and speed measurement) have been improved greatly since early navigation days, with corresponding increases of position accuracy.
Acoustic navigation¶
The robot has another valuable source of navigation from the acoustic pinger deployed as a part of the competition. The regular timing and fixed location of the signal source could make it positioning tool. This is known as baseline navigation. There are a number of established challenges to baseline navigation, including acoustic multi-path and the angular resolution of the receiving array. These challenges mean that the acoustics will likely only be used to get additionally points in the competition, and not for navigation until there is significant further development.
Competition strategy¶
Focusing on dead-reckoning navigation, two measurements are required for accurate positioning; heading and velocity. Both of these measurements are available to the robot in two flavors, relative and absolute. Relative quantities are found by integrating an inertial measurement, angular velocity and acceleration, respectively. Inertial measurement have relatively good precision, but their error increases with integration time. An important consideration when using these measurements is the amount of time over which the integration can be considered accurate, after which time it should be restarted.
Given that inertial measurements are only valid for a length of time, it is best to use absolute measurements. These are taken directly, in our case using a compass and a Doppler velocity log (DVL). There are challenges with both of these measurements which require the mention of the relative measurements in the first place. As these effects are further studied and addressed, perhaps it will not be necessary to use relative measurements further. This would significantly clarify the navigation of Zoidberg.
The compass is influenced heavily by the internal magnetic field of Zoidberg, including permanent effects of iron and transient fields caused by large currents in motor power cables. Additionally, the external magnetic field inside the competition pool is corrupted by the large amount of iron used in its construction. This effect is expected to be strongest near the walls and basement of the tank. The current strategy for absolute heading measurement will be to minimize internal interference with the compass, and work to identify places in the tank when the heading measurement is trusted.
The DVL is a relatively untested system on Zoidberg because it does not seem to work well in small pools. This is marked by a great deal of dropped measurements, which then require a back-up velocity measurement system. The simplest way to do this is to measure the relationship between motor input and robot speed. If this is well behaved (more power gives more speed, mostly linearly) then it will be possible to predict robot speed from motor input.