There are different many reasons why we need to build something physical. Some examples are: proof of concept, demonstration or presentation of appearance or shape, show a certain functionality, performance measurement, producibility, etc. Depending on the needs, we can decide how realistically it has to be and we might build the part to proof realistically and use stubs or old or simplified versions for the rest.
To build whatever it is we need, we first need the materials, either standard (of-the-shelve) or custom made parts. In case of standard parts, it helps if the most common parts are readily available. You can either order these parts in advance (keep a minimum stock level), tap into reverse logistics (left overs from customer sites or old tests), use factory stock, or other means. Depending on how often something is needed, you want to make it faster and easier to retrieve.
There is of course a limit to what you want to keep on stock. In case the standard part is not on stock or you need something custom build, make sure the ordering process is fast. Some ideas for this are: have a list of preferred suppliers that developers can contact and order directly using a sketch and without going through a length purchase order process, order via a web shop by credit card or on account, use a courier to deliver or pick up yourself, use <24hr delivery shops, automate order and supply processes, etc. To find options for improvement, a value stream map is a great tool to use. Just write down the process steps from order request to delivery and see what improvements can be made (speed up step, alternative way, skip step, …).
A third way to speed up when we need a part that we can manufacture ourselves. You need a workshop at hand where you can manufacture different parts, make minor adjustments, assemble, etc. An emerging concept is the desktop factory.
This means having all kinds of desktop size tools like a 3D printer, CNC machining, vacuum forming, etc. When you need something bigger or unable to make yourself, a workshop nearby would be the next option. Some bigger companies have special workshops that you can make use of or you can use rapid protoshops in the area or even further away as long as they can deliver fast. For most parts you should get a <24 hour delivery from order to desk.
The fastest way to test rapidly is usually virtually. For examples, please see Model Based Engineering. When a physical test needs to be performed, there are many creative ways to automate physical testing with varying budgets. See below for some examples. What also helps a lot is to design for agile testing. This means making sure you can quickly replace something that needs to be tested often (click on/off, plug and play, magnetic), manufacture parts with test probes or sensors or extra holes or slotted holes or …, have 2 versions (e.g. via feature toggles): one for testing and one for normal use, etc.
How about some automated testing of suspension at Red Bull F1?
How about using a scaled model? Or, how about using a simplified model?
How about a robot automatically touching the screen?
How about this “carnival ride” with a serious touch.
How about robots working together to verify producibility or to create the product and then automatically run a series of tests on it?
How about a pick & place robot to continuously run tests?
How about a pick & place AGV to transport your test items to the test area for the next test.
How about using realistic environment conditions, even as extreme as in space using the Large Space Simulator at ESA?
Or, robot assistant? Flexible to do complex tasks, doesn’t get tired or bored.