-- MarkCutkosky - 29 Jun 2004

Emails from CMU with instructions

Setting up the new leg in Stanford
June 29, 2004 3:02:49 PM PDT
gch@cs.cmu.edu, hkomsuog@umich.edu, aaron@BostonDynamics.com, dcampb@BostonDynamics.com, arizzi+@cs.cmu.edu, cutkosky@stanford.edu


We should definitely have a small telecon later in the week to straighten things up and address any questions you may have. However, here is a summary of the steps I think you guys will need to go through to get the new system up and running.

First, brief descriptions of all the components that are involved with the new setup:

- The RMC card: This is the H shaped card that is densely populated that we sent you last week. This card was designed by Don Campbell and is responsible from electrically interfacing the motors, encoders and hall effect sensors. This card has two MCU processors (one for each hip) that includes firmware to access all these hardware components as well as implement the "RiSEBus" protocol for communicating with the CPU. Don has documentation for the specifications and schematics of this card, and I am sure he would be happy to send them to you assuming you are covered under the NDA.

- The RiSEBus? bridge: That is the "yellow" PC104 style card that we sent last week. This card was designed by Hal Komsuoglu and it is used to establish communication between the main CPU and the RMC cards. It has four "bus" channels, although the one that we sent you is only populated with one channel. Normally pairs of hips (front, middle and back) and the body peripheral devices are connected to one channel each. In your case, because you only have one hip module, you will only need one channel. If you look at the card, you will see two connectors close-by at one side of the card (one 10 pin and one 6 pin). The 10-pin connector is for programming the MCU firmware, which you do not need to worry about. The 6 pin connector is for the cable that dangles off the RMC board labeled as RiSEBus? . This cable is where all the communication with the hip module goes through. You will probably need to build a small extension. Hal can provide you with details on part numbers and documentation for this bridge card.

- The stack that you will need to build is somewhat simpler than the one that you are using right now. It should have the following cards:

1. Power supply. Takes 24V and powers the PC104 stack 2. CPU card. The Lippert card that you are using with the current stack should do the trick. Also, the current QNX installation on the flash should be ok. 3. PCMCIA carrier. Even though it is possible to use the onboard ethernet of the CPU for network connection, it may be more convenient to use the wireless card with peer-to-peer operation. I suspect that's what you guys are doing with your current stack, so you can keep his as it is. 4. RiSEBus? bridge.

You will need to play around with the order of the cards to figure out what fits together best without the card components interfering with each other.

- I believe this is most of what you need to put the hardware together. The RMC documentation should tell you how to connect the motors to the RMC card. The next bit you will need to tackle is the software. I recall that you were using the RHex software for controlling the test track. You will need to get up and running with the real RiSE software, which is located under RoboDevel? /RiSE. It is also RHexLib based, so you should not have any trouble with most of the concepts like modules, modes etc. The hardware library under RiSE/RobotCode/Hardware/RiSE_RBHW does all that is necessary to access the RMC cards and control the motors through the RiSEBus? . This will take a little bit of effort to understand and maybe deal with small issues that may arise from the abscence of 5 of the hip modules.

- One important issue that needs to be tackled is calibration. I had not had the chance to look at the hip assembly Aaron sent you, but I suspect it does not have the hall effect sensors for the wing degree of freedom. This eliminates the possibility of using the calibration procedure we use for the actual robot. As a consequence, there are two things that can be done. 1. build the hall effect sensor assembly, 2. Come up with an alternative way to calibrate the hip and code up an alternative module for calibration. I would recommend the first option as I think it is easier to implement, it will make experimentation much simpler and you can use the same pieces of software to perform calibration.

That's all I can think of right now. Once you progress along these lines a little bit, we can flesh out some of these and address questions as they arise. Let me know if any of this did not make sense.

Good luck. Uluc.

On Fri, 25 Jun 2004, Arthur Joseph McClung? wrote:

Hi Clark,

Very exciting work presented today! Its starting to look more and more like we have a legit project. I think I remember hearing that wing angles above 75 may cause joint conflicts, thus the robot climbs with legs at 45deg and returns the legs at 30 deg more (ie. 75deg). I just wanted to verify before we begin any testing on the new leg, and put some safety checks in the interface that easily creates RC files for CustomGait? use.

Also, I was hoping to get the new leg up and running sometime next week, so I was wondering what might be the best way to get some advice on the hardware setup (email or should I try to schedule a phone call to you/Uluc/Aaron/other about getting everything in the right place).

Thanks again for all the help, Trey

Uluc Saranli Postdoc, Robotics Institute http://www.cs.cmu.edu/~ulucs Carnegie Mellon University phone : (412) 268 3260 5000 Forbes Avenue fax : (412) 268 5571 Pittsburgh, PA 15213 email : ulucs+@cs.cmu.edu

This site is powered by the TWiki collaboration platformCopyright &© by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback