Monday, September 29, 2008

Legs almost ready

Thursday, September 25, 2008

Mechanical assembly started!

Most of the mechanical components are ready, so assembly has started on last Tuesday. Most of the time has been spent assembling oblique bearings due to interference. One foot has been completed and has been tested on soil. Also one ankle has been almost completely assembled. It is still the old design of the ankle, as the components of the new one are expected to arrive next week. There is still a lot of work to do, but as we manage to complete the first leg the other ones will take much less time.

Friday, September 19, 2008

The Communication System

As we promised, here is the detailed status of our multi-hop communication system:
  • The low-level control FPGA boards are ready. They implement the PIDs ensuring smooth operation of drivers and sensor readings from the encoders. We devised a flexible protocol to send commands to the motors and get timely sensor readings to and from both the on-board computer and the mid-level control (pattern-generator) board. We used Altera DE2-70 boards, communication takes place over Ethernet.
  • With a well-defined protocol we can now begin porting the Matlab pattern-generator code to the FPGA board which will ensure real-time performance.
  • We are coding the software which will run on the on-board computer. Our current focus is getting good performance out of the stereocamera and delivering a reasonably good (and packet-loss ready) video stream. We are still having some difficulties with buggy libraries and fuzzy-behaving encoding software, had to write to PTgray a couple of times. We have our Linux system running on the Fanless VIA EPIA, we tested all basic interactions with the various components.
  • Good news from the Multi-hop communication system: our Fonera nodes are up and running with B.A.T.M.A.N., a robust multi-hop communication system. Testing shows very good range (with some hops we can communicate from side to side of our large and thick-walled historical building with only 10% packet loss) and a comfortable bandwidth of more than 500 kB/s. The home-made battery pack got us 2.5 hours of run-time in little weight.
As we reported in our previous post, we are still waiting for some equipment. As more and more subsystems get ready, our tests are becoming increasingly realistic. In a short time we'll run a complete motor tele-operation test.

Wednesday, September 17, 2008

Mechanical design and assembly update

It has been a pretty long time since the last detailed update on mechanical design and robot assembly. It has not been a summer of relax and vacations, though! The mechanical design team has been working for the whole summer to turn the ideas we presented at CDR into feasible models and drawings that could be sent to the factories.

During this phase new technological and practical problems have been arising, so we had to perform several changes to the preliminary design. The main guidelines that lead our work were to reduce the weight as much as possible and to keep the manufacturing and assembly process as simple as possible. The time constraint has been very important, too. We avoided components that could not be delivered in time. Hereafter is a summary of the major improvements and changes.
The truss-like frame has been substituted by a carbon fiber box made stiffer by several reinforcements. The ankles have been completely re-designed to reduce encumbrance and weight and to increase the workspace. With the new design two of the three motors that power each leg will no longer be mounted in a vertical position, but they will be horizontal allowing a further reduction of complexity (conic gears are no longer required). Feet have been revolutionized, too. They are now smaller and simpler to assembly since we have greatly reduced the number of components. They also include a removable spike that can be used on less firm soils.

By now, all the components made in composite have been completed by our supplier C.R.M. Compositi in Livorno and also many of the steel components are ready. Everything is being transferred to the Micromecc factory in Ospedaletto where we will soon begin the assembly phase.

Status update

Long story short: things are going well, but we still have a lot to do.

This month we have completed our Fonera-based multi-hop wireless system and we did some tests on the wireless node battery packs obtaining a satisfactory 2.5 hours uptime. We got our hands on the IMU, the spotlights and the cameras, which we got working on Linux with our streaming software (still to be completed). We have had some problems with the main battery pack shipment but the new one will arrive before the end of the month.

Mechanical parts are arriving from manufacturers and we hope to begin assembly next week. Motors have been shipped two weeks ago and low level control boards are operational and accepting commands from ethernet. Right now we are working on some communication issues between the three different control levels, and we expect to solve them in little time.

In a few days we will be ready to begin porting the pattern generator code from the Matlab dynamics simulator to the middle-level control FPGA board. We should be able to perform our first full motor control test on Monday.

Altera kindly donated us three DE2-70 FPGA boards, on which we implemented both the low and middle level control.

We also have started porting the communication and image capture software on the onboard computer (based on a VIA EPIA fanless board). Moreover, Dell has very kindly offered to donate the teleoperation station laptop, we'll get in touch shortly.

Thanks to everyone for their hard work, in the next few days we'll post the detailed status of each subsystem.