We now have the breadboard prototypes of almost all the components up and running on our test bench. We are running them through multiple different types of tests. Initially we are running them unloaded over their range of motion to make sure there are not clearance or fatigue issues. Our tests continue with adding weights the size of the maximum rated load. And then we are extending the driven range to hit the hard bump stops and test the whole system in the worst case scenario. The picture is our test setup in the lab.
More information on the test setup is on the wiki.
The two prototype pieces for our sensor head arrived yesterday. As a software guy I get sit back and say, "that's cool," and not have to fully appreciate the work that went into it. What impressed me most is how much thought is going into making sure that it meets our needs yet remains hackable for other researchers: there's plenty of room to add your own equipment and you can easily remove what's already there.
The photo on top is the large pan-tilt platform with Axis 213s mounted on the side. It's hard to tell from the photo, but it's big -- wider than your shoulders, probably. We're planning on mounting stereo cameras to the pan-tilt, but there's an entire grid of bolt pattern on top to customize with. The photo below is the tilting platform for the Hokuyo that just barely squeezes underneath. Keenan claims you could run the tilting platform at 10hz, but we probably won't test that.
We'll be testing these hard to figure what we need to stiffen, add, remove, shave, or otherwise improve upon before final release. It also gives us some more hardware to write code against. I didn't get a chance to take a nicer picture of the titling platform because we're already busy breaking it in.
This week we released designs for mechanical breadboards of all the system components (base, torso, arms, forearms, hands, and sensor head) to manufacturing. Getting components made will take about a month, so we'll have more mechanical components coming in throughout April. These are the tests to make sure all the subsystems are ready for us to build up an Alpha run of complete robots in June.
We have the a first preview version of the ROS architecture publicly available. It's a long way from the complete system, but it's enough to get a basic distributed system up and running.
Documentation and quickstart instructions are on the wiki
We have the first real mechanical components of PR2 assembled - the steerable driven casters that move it around. These were built by Function Engineering, and we're in the process of testing and building them together into a base. The sensor with three wires hanging off is a hall-effect sensor for absolute calibration of rotation. The caster also has three motors and three incremental encoders - one to rotate the caster and one to drive each wheel.
