Faculty Representative for the
United States Naval Academy's
Autonomous Underwater Vehicle Club

This project and page are both works in progress.

AUV mockup in tow tank for drag testing.

Tow Tank Testing Movie (Quicktime 5 MB)

AUV Configuration and Programming Guide (PDF 150 KB).
Work in Progress

Contents

1. Project Background
2. Vehicle Functional Requirements
3. Configuration Ideas
4. CPU Ideas
5. Hull Ideas... The Tyranny of the PVC Pipe Idea
6. CPU Administration
7. Control Surfaces
8. Depth Cell

Project Background

Being the faculty representative for a project like this is tricky. You want the students to do as much of the work as they can and at the same time you want to steer them away from the really bad ideas. Additionally, in a project as complicated as this, there will be some aspects which are just too hard for the students to do in the time they have allotted for the project.

This club was formed with the intention of building a vehicle to compete in AUVSI's Student AUV Contest.

Before the student's got involved, the other faculty representative and I felt it necessary to select and become familiar with most of the major components for the vehicle. This allowed us to provide some good instruction for the students on interfacing with the hardware. We selected the following components:

• Main CPU
• Main Thrusters
• Hull Enclosure
• Doppler Velocimeter Log
• Compass

Vehicle Functional Requirements

One of the first steps that we helped the students work through was determining the functional requirements (FR) of the vehicle. These FR's were determined from two sources: the rules of the AUVSI Contest and other envisioned uses for the vehicle.

Contest Imposed or Derived FR's

1. Minimum Weight: We set a goal of less than 40lbs (There is a point penalty for heaver vehicles).
2. Able to drop a 1/2 lb weight at a particular location
3. Able to home in on and touch a light source with particular characteristics.
4. Able to follow a PVC pipe on the bottom of the pool
5. Maximum depth 25 feet.
6. Perform 3 contest attempts in 15 minutes
7. Swim straight through a validation gate
8. External kill switch which cuts all power to the vehicle.
9. 5 percent positively buoyant.

   Additional FR's

10. Able to wirelessly communicate with and reprogram the vehicle.
11. Able to add payload modules for future student projects.

Configuration Ideas

We had the students think of many different ways that thrusters and control surfaces could be arranged to maneuver a vehicle in three dimensions. Here is an example of three configuration considered.

Thruster Configuration 1

Port and starboard main propulsion thrusters. Fore and aft vertical thrusters for depth and pitch control.

This configuration is one of the more common configurations seen at the AUVSI competition. The only direction this vehicle cannot travel is sideways.

Thruster Configuration 2

Three main propulsion thrusters arranged in a triangle (when viewed from stern).

This design uses only three thrusters, thereby saving weight and money. All three thrusters are main propulsion thrusters. By operating the thrusters in a coordinated fashion they can control both pitch and yaw. Stationary depth holding is more difficult. All the thrusters can be used for forward propulsion, so this is a more efficient use of thruster mass. By comparison, configuration 1 only uses 50% of its thruster mass for forward propulsion.

Thruster Configuration 3

This configuration uses a single main thruster for propulsion. Control surfaces are used for pitch and heading control.

This configuration only uses a single main thruster for propulsion. Pitch and heading are controlled by control surfaces.

Advantages	Disadvantages
Likely to have the lowest power consumption because there is only 1 thruster. Likely to be the smallest possible vehicle.	Vehicle must be in motion to control depth, heading, and pitch. Difficult to hover Control surface linkages make inefficient use of interior volume and provide more leakage paths.

Other Thruster Configurations

Many other thruster configurations were considered. Including:

• Port and starboard main thrusters and a single vertical thruster for depth control.
• Port and starboard main thrusters which can pivot 90 degrees for depth control (like a Harrier aircraft).
• A single main thruster with a movable nozzle allowing the AUV to control pitch and yaw.
• and many others...

Configuration Evaluation

Then I took the students through the exercise of selecting criterion for comparing each concept. Below are the results from one round of comparisons.

The rankings for configurations 1, 2, and 3. This round of comparisons indicate that configuration 3 is preferable to configurations 1 or 2. See the text for why we chose to NOT use configuration 3.

The above figure shows a set of comparisons which the students made. This comparisons indicate that configuration 3 is preferable to configurations 1 or 2. There we a couple of problems with this comparison.

• The weighting assigned to ease of manufacture is way too low compared to the other criterion.
• Estimating configuration 3 as easier to manufacture than configurations 1 or 2 is questionable. Configuration 3 requires at least 2 sets of control surfaces. Each of these control surfaces requires a linkage and shaft seal. The control surface/linkage/shaft seal combination would likely have to be custom designed and installed. Additional thrusters only require additional waterproof electrical connectors which are easy to install (not cheap, but easy).
• Additionally configuration 1 is much more adaptable than configuration 3 because it is less sensitive to ballasting. This allows the thrusters to be moved to allow the installation of additional equipment. In an extreme case the thrusters/CPU/Batteries can be moved to a completely different frame without changing the basic vehicle configuration.

The final configuration for the first version of the vehicle.

The final thruster configuration for the first version of the vehicle is not one of the configurations listed above. In an attempt to save weight we decided to use a port and starboard main thruster and control surfaces to control depth. Personnel usage was another factor in deciding to try to build a set of control surfaces. During the semester there were 12 students working on the project. Designing and building the control surfaces and their actuators in a waterproof housing was a full time project for three of them.

The control surface project was a chance for those students to take a project from the initial concept to the final detailed design through to the construction. This was a fantastic learning experience which is the point of the exercise of course.

As it turns out, building a waterproof box that is lighter than a pair of small commercially available thrusters is harder than they thought. It is likely that the control surfaces will be replaced with a pair of thrusters for depth control.

CPU Ideas

The table below lists the components we wanted the CPU to interface with:

A number of factors went into selecting the CPU for the vehicle. These included:

• Power Consumption / Heat Generation
• Wireless Configuration / Reprogramming
• Powerful enough to perform the required image and sonar processing
• Students familiarity with the programming environment
• Cost
• The number of additional components needed to interface with onboard equipment.

*Running the computer from onboard flash drastically speeds up boot times (less than 5 seconds) and provides a journaling file system (which handles inadvertent loss of power much more gracefully). But the gcc compiler and development environment will not fit on the onboard flash so compiling must be done off the board.

I selected Technologic Systems TS-7200 for the AUV CPU. I continue to be very impressed by this board. It is an excellent choice for projects like this.

Hull Ideas... The Tyranny of the PVC Pipe

"Lets just get a piece of PVC pipe and make the hull out of that!"

Anyone who has every tried to build an AUV for the first time

The only benefit of the PVC pipe hull is hydrodynamic.

Here are the problems with a PVC pipe hull:

• It is very difficult to see inside the vehicle without significant disassembly.
• PVC pipes aren't very round. Getting a good seal on the end cap typically requires turning the sealing surfaces of both the pipe and end cap. This can be particularly difficult if the pipe (a.k.a. hull) has a large diameter and length therefore is too big for the lathe.
• The guts of the vehicle need to able to be removed from the hull as a single unit. This has a couple of follow-on implications:
  • All the through-hull electrical connections need to be made on the ends of pipe. Otherwise, you can't slide the interior of the vehicle out of the pipe.
  • Holes for the electrical connections will make the end cap look like swiss cheese, requiring it to be even thicker to maintain adequate strength.
  • It is very difficult to make an end cap that:
    1. is removable
    2. is waterproof
    3. does not require large flanges (thereby eliminating some of the hydrodynamic benefits of the pipe hull)
    4. aligns the exact same way when installed
    5. requires no rotation to install or remove
  • This means that one end of your pipe will likely have a fixed end cap with the electrical penetrations and a removable end cap which allows access to the vehicle. But this means that there needs be some sort of back plane connector on the inside of the fixed end cap to allow the interior components of the vehicle to be removed.
• Some external structure is typically required to mount other components which don't fit inside the hull. This reduces the hydrodynamic benefits of using the PVC pipe for your hull. There have been vehicles built where all the components are internal to the pipe. This eliminates many of the above problems. However, these vehicles are harder to design.

The above reasons are why I selected Pelican Cases to provide the waterproof enclosures for the vehicle. They provide:

• A full size opening allowing complete access to the inside of the vehicle.
• 6 flat surfaces to mount electrical penetrations on.
• They are readily available.
• Little additional work is required to make them suitable for our use (sand and drill some holes).

CPU Administration

I have started writing an AUV Programming Guide (PDF). Currently, it covers the initial configuration of our ARM based single board computer. This too is a work in progress.

Control Surfaces

We decided to try to build control surfaces to control the vehicle's pitch and depth. For this to be worth while the control surfaces and their actuators should weigh less than a pair of thrusters. Otherwise it is simpler to just use a pair of thrusters for depth control. Additionally, this was undertaken to give three of the students a small self contained project which would take them all the way through the design process.

"Its a waterproof box with a motor inside and shaft coming out.
How hard can that be?"

A student to remain unnamed

The basic vision was a waterproof box with the shaft for the control surfaces coming out either side. This "servo box" could then be mounted on the AUV's frame in the desired location.

Different types of actuators were considered, including stepper motors and hobby servos. The students selected hobby servos as we already had circuit cards to control them via serial commands.

Once servos were selected, the students considered several mechanisms to connect the servo to the shaft. These included linkages, gears and hydraulic rams. The students elected to use a linkage.

An early sketch model of the servo box.

An early sketch of the proposed shaft seals.

The final servo box was a folded and welded sheet metal box. The shaft seals were tapered cones and o-rings. The there was a lot of wasted space in the box. Therefore, weight had to be added to the vehicle to make it neutrally buoyant. This causes the servo box to weigh much more than two thrusters. Additionally the linkage connecting the servo to the shaft is too flimsy.

Here are some photos of the final servo box for the control surfaces:

Depth Cell

Initially the students tried to use a depth cell from the past vehicle. This cell required an external bridge circuit and regulated power supply. Additionally, since the cell was designed to be mounted in a pipe the end with the wires was not waterproof. So it had to be potted in epoxy, since the depth cell on our vehicle would be mounted on the frame with an external cable running back to the CPU housing. Calibrating the bridge circuit also required making a 30 foot water column.

Calibrating the old depth cell.

This old depth cell was replaced by a fully waterproof cell which can take 13-28 VDC, and provide a linear 0-10 VDC proportional to 0-30 feet of depth in a single compact package requiring no external circuitry.

The new completely self contained depth cell

This project and page are both works in progress.