As several people have advocated using pressure vessels, I'll add my own take on this.
Pressure vessel safety
Compressed air in an uncertified vessel can be extremely dangerous. You need a pressure of about 1.2atm gauge to simulate 40ft of water. That's about 2.2atm absolute. I'm not going to perform an integration here, but air at these pressures is compressed to about half its original volume and contains about 120J of energy per litre, enough to accelerate a 100g projectile to 50m/s (180km/h,110mph.) Water on the other hand hardly compresses at all when pressurized to these pressures, and therefore hardly expands at all when it depressurizes, so the energy release in case of a rupture will be safe.
It is normal practice before commissioning a pressure vessel to fill it completely with water, then pressurize it, in order to test its integrity (a "hydrotest"). On very rare occasions where this is absolutely impractical, a very thorough inspection of the welds (radiography, dye penetrant, ultrasound, magnetic permeability) and joints may be done before testing with air. Neverthless, professionals prefer to test vessels with water in order to avoid the risk of a pressure explosion. In my opinion, amateurs should take the same precautions.
For anyone who thinks this is not a big issue, this is what happens when a small beer bottle explodes. https://www.youtube.com/watch?v=JqXh86ks_As
Note that engineers do sometimes get things wrong. Here's an example where it was impractical to perform a hydrotest. They thought they had a safety factor of 2.5 on pressure, but there was a weak point that they overlooked. https://en.wikipedia.org/wiki/BOAC_Flight_781 . Note that in the accident investigation they did perform hydrotests on an aircraft fuselage, until destruction. To perform the same tests with air would have been extremely dangerous.
If your robot (or its components) are small enough, you can use a pressure cooker. It is quite likely that you can get 17-18psi from the mains water in your area. If not you can use a central heating pump to boost it. If you are sure the pressure cooker is rated for the required pressure you can ignore the next paragraph.
Your pressure cooker likely has only one valve connection, which is not centred. This will make it difficult for you to expel the air once it is sealed. So immerse the pressure cooker in a bath, put the robot inside and put the lid on. This will ensure no trapped air. Pressurize with water from the mains or a pump. After use, do not use the pressure cooker again as a pressure cooker unless you are sure you have not exceeded its rated pressure.
If you manage to acquire a larger vessel, it's even more important to ensure all air is expelled. Try to arrange for the filling point to be at the top. If you have a hose of smaller diameter than the filling point, air can pass out around it until the vessel is full and water starts to pour out. Again mains water pressure or a small pump will be required.
The sidewalls (say 2mm thick) of an oil drum (say 400mm dia) would experience a stress of 17psi x (400mm/(2x2mm)) = 17000psi and should in theory hold (but I'm NOT saying they would be in code.) The top and bottom would however become severely domed. I would recommend something thicker, like Sch40 pipe, with some decent ends.
You can monitor your pressure with a tyre pressure gauge, or control it with a 40ft high overflow pipe (if your building is high enough this is easier because you can leave the flow of water on and not worry about exceeding pressure.)
SkipBerne's suggestion of a fibreglass pool filter seems an excellent idea, much larger than a pressure cooker. http://www.pentairpool.com/pdfs/FNSPlusOM.pdf Seems this model will take 50psi. Note text in the manual "AIR ENTERING YOUR FILTER IS DANGEROUS" (capitalization theirs.)
You can simulate a pressure differential of 1 atm (14.7psi) by fitting a valve to your robot and drawing the air out of it. That's nearly 17psi. Given the possibility that the vacuum pump will have to handle water if something goes wrong, I would not recommend using your physics lab's best vacuum pump. Go to the chemistry lab and get a cheap eductor type vacuum pump. These use a jet of water and the venturi effect to produce a vacuum for removal of air from dessicators. If you dye the water surrounding your robot, you will be able to detect any breach by seeing the colour in the hose to the vacuum pump.
You may find finally that the cheapest option is to find a reasonably deep lake (possibly a reservoir with a dam, they tend to be deep) and try it out for real there. They may not like you hanging your robot off a rope at the dam, but many reservoirs are used for boating so you could hire a boat and do it in the middle somewhere. Obviously don't offend the relevant authorities. Failure to recover your robot, besides being a setback for you, would be unauthorised waste disposal. Be careful with toxic substances such as batteries.