This post originated from an RSS feed registered with Agile Buzz
by James Robertson.
Original Post: Autonomous Underwater Vehicle
Feed Title: Michael Lucas-Smith
Feed URL: http://www.michaellucassmith.com/site.atom
Feed Description: Smalltalk and my misinterpretations of life
Back in 1997 doing underwater vehicles would be interesting to do. So he taught himself electronics and micro controllers, etc. In the last few years he's actually had a functional AUV.
Or "Unmanned Underwater Vehicle". It's a submarine, it's submursble. This model doesn't have a tether, this is completely wireless. He can't get instant wireless back but he has complete mobility.
Everything is onboard, batteries, sensors, etc. They're used for lots of things. Science, environmental monitoring, undersea survey, cable/pipeline inspection. Various military uses - including mine detection and clearing.
Why are they difficult to build? They're truly autonomous, fault tolernce is critical. Hard to get real-time feedback. Comparing to the Mars Rover mission because they control themselves. We have to wait to get them to do something. The AUV is worse, once it goes underwater it's hard to get feedback from it. The board rates are very very slow - acustic modems. It's very unreliable. They have very limited senses under water.
RF doesn't work under water. You can't use GPS, radio, etc. Light is very difficult to use under water. Salt water blocks lots of the light. Therefore, they have to rely on sound a great deal - sonar pings. Water pressure as well to know depth. The pressure increases the deeper it goes.
AUV's have to know where they are. On land, in air we use GPS. But under water that cannot work.
They're difficult to build because it's a dangerous environment - wet, corrosive, pressure increases with depth, expensive engineering, failures are often catastrophic/fatal - you lose it and can't fetch it back. Salt water is especially corrosive.
The cost goes up linearly or more depending on how deep you want it to go because of engineering and building materials.
Commercial AUV's come from places like Bluefin, ODyssey II, Mainly survey vehicle because most companies make ALV.
Question: Why are these commercial ones so big? THat big one on the screen there is actually a small one. You need a serious crane to pick the larger ones up. Thus you need a decently sized boat.
ISE - International Submarines 'E'somethingerother make Theseus which is 35 feet long and it is designed to lay fibre optic cable. It can go 1000 metres deep. It lays cable between arctic research centers. Theseus was the greek god that got lost in a maze and left thread behind him to find his way out.
Konsberg Simrad make the Hugin AUV which can go 3000 metres deep. That covers most of the ocean. It's for surveying.
Hydroid, Inc make Remus is a tiny AUV about 5 feet long. Very cheap (as in: $250,000) and the Navy uses them to clear mines from Umm Qasr in Iraq. These are to replace the dolphins they were using.
MicroSeeker is this guys AUV, it's cheap at $5,000. It's very small only 24" long. It has a depth of 30 feet only. But it is fully autonomous with RF connection to base. Acustic modems cast even more than the whole AUV, so he just uses the RF. The main computer in it is a PDA that has a wifi card which lets him talk back to his computer.
His old MicroSeeker looked like the Enterprise, but it didn't work so well. The thrusters didn't work well enough - couldn't navigate.
Two weeks ago they did a pool test with their latest model and it actually survived! :)
A video was shown of the test run where they made sure the thing didn't leak, that it could go in a straight line and that it had correct boyency. But the engine didn't have enough torque to navigate around the pool successfully.
The batteries don't have enough "juice". Solution, replace the motor with a 12 volt gearmotor. Replace the dual 7.2 volt AA battery packs with a single sealed lead acid 12 volt battery. Need a new circuit board - lessons learnt from last one, plus new power requirements.
This AUV literally has a PDA sitting in its center. The body is made of glass so you can see all the innards.
MicroSeekers 'brain' that makes it "do" things is written in Squeak running on the PDA and communicating back to the main computer for new commands. The machine is 300mhz, 32mb of ram, flash, wifi, etc.
As soon as the device goes under the water, the wifi connection is lost, so all commands must be given to it before it dives.
Smalltalk makes the programming, the complex programming, much much easier.
There is an abstraction between the mission layer - the goals, mapping, routes, decision-makers based on sensor feedback.
I should point at this point that he hand tools all of the materials used in this system as well as programming Squeak to do all the communications and mapping, etc.
Sensors on this sub: speed sensor (from paddle-wheel). Heading, pitch, roll (from 3-axis compass). Depth (from pressure snsors). Bottom Sonar (not completed yet). Leak Status (spring, aspirin tablet, contacts). Internal status (software).
The leak status, there is an aspirin tablet on the bottom of the sub with a spring on it. If water gets in the tablet dissolves and the spring makes a contact which tells the sub to activate an abort immediately.
The sub uses a Navigator class which defines the rules it's going to follow in a mission. There may be a PingNavigator which steers the sub towards the ping that it's hearing, for example. Basic ones just keep a depth, a speed and a bearing.
He also has a simulator for the sub written in Squeak - so that it's all live. This was brilliant. He added in some variation because the sensors in real life are not perfect and the water you're in may be rough.
Because it is all Squeak, all the objects are live objects that he can interact with. In this demonstration, the sub then sets up some way points to do a loop near the beaken that it has driven too.
Even in the simulation, it has gotten lost because the model of the sub is unable to turn fast enough to get to a waypoint. Because it cannot get to the way point well enough, it continues to attempt to meet the waypoint and logs that it's failed.
Where to go from here? The hardware improvements to keep the sub under the water and be able to turn. The new controller will be added and the next AUV design is a 3 feet long 10" diameter AUV with a 100 metre depth capability. Single vectored thruster with 50-60 hours endurance. And running Squeak of course :)
The new AUV looks very cool, it looks like a torpedo instead of an Enterprise shuttle (the first looked like the Enterprise).
His company is Hylands Underwater Vehicles. The dream is to make the company use smalltalk to solve real-world AUV problems. Immediate goal is to get funding to build the Seeker. The Seeker will cost around $60,000 (in parts) to build.
Previous Topic
