Parallax Insider News

University of Tennessee Graduate Robotics Team

  • By: Parallax Insider Published: 18 December, 2013 0 comments
The IEEE Graduate Robotics Team at the University of Tennessee, Knoxville, is a self-organized group of graduate students who seek to participate in major robotics competitions.  In April 2013, our team won 1st Place in the IEEE SoutheastCon 2013 Open Hardware Competition hosted in Jacksonville, FL.  This competition tasked students with designing a robot that could complete a loading dock delivery task.  This required the robot to perform tasks relating to image processing, navigation, and more.  We used a number of Parallax products as part of our entry in this robotics competition and found the products offered by Parallax to be very instrumental in the success of our robot.
 
When we first looked into robot design for the competition, we knew that a pre-made chassis would offer the quickest and easiest option to start both our hardware and software teams working in parallel.  The chassis could offer the software team a development platform to work on movement code while the hardware team could begin adding customized parts to better fit the competition rules.
 
Since the competition layout included an inclined portion and the overall construction slop was about an inch, we decided that we needed to focus on traction and carrying capacity as important design focuses over speed or small size.  We also tried to find a design with built-in encoders, which could be used again for the basic navigation code.
 
The Dagu Rover 5 Tracked Robot chassis worked perfectly for these tasks - four wheel drive and an adjustable height (we used ours at the lowest position, as our overall height was potentially an issue) really helped to prevent slippage as we tested driving up in the inclined ramp.  The basic measurements of the mounting locations (located in a rough rectangle on the top of the chassis) allowed us to 3D-print an adapter plate that provided for easy access to the chassis itself if we needed to disassemble our robot.
 
A few modifications were made to the chassis that might help future projects.  To help facilitate linear movement of a claw mechanism, we mounted a large threaded standoff in-between two of the drive motor "legs".  There's plenty of room (once you disassemble the legs) for mounting, and the extra rigidity helped the chassis to track completely straight over long distances.  Also, we found that the chassis worked well with a much larger battery pack (NiMH) as carrying capacity was no issue.  The chassis, having a low center of gravity, has no problems with a battery pack mount on the front or rear vertical surfaces.
 
After doing some navigation tests using the tank treads and servo encoders on the pre-built chassis, we quickly realized that the combination of the tank treads and encoders posed a significant problem given our application and goals.  In order to use the tank treads to turn the robot, it was necessary for the treads to move in opposing directions.  This caused the treads to slip and thus introduced inaccuracies in the encoder readings.  These inaccuracies made it very difficult to determine the magnitude of the robot's turns and therefore made it difficult to achieve the precise movement control needed for our application.
 
In order to give the robot a greater degree of precision in its turns, we mounted a Parallax 3-Axis L3G4200D Gyroscope Module to the robot and used it to measure angular rotation of our robot as it turned.  The addition of this module allowed us to vastly improve the turning accuracy of our robot by allowing us to directly measure the turning rate instead of having to guess the turn angle based on unreliable encoder readings.  Once the new gyroscope module was installed on the robot and the navigation code was adapted to read measurements from the gyroscope, the turning accuracy of the robot improved vastly.  With this improved turning accuracy, we were able to fine-tune our navigation system and achieve unprecedented performance in our competition runs.
 
For wireless communication, our team has used XBee radios from Parallax for the past two years now, and has yet to find a better wireless solution.  The problem with any wireless communication system usually focuses on a complicated setup process - syncing devices, matching transmit speeds, programming automatic "ack" broadcasts, etc.  The XBee system takes care of all these issues automatically - you can't get any closer to plug-and-play than using a set of these.
 
We particularly like XBees for their debug and testing usefulness.  An XBee connected directly to a microcontroller can be connected into a standard USB port, then used for serial communications (to a second XBee/microcontroller module).  In terms of programming, writing data to a COM port and waiting for a response is very straightforward and error-free.
 
We used a set of XBee radios this year for communication between a base station that provided all the heavy image processing and a mobile robot that performed various required tasks.  Any time we needed to debug this wireless interface, a third XBee/microcontroller module and a telnet client such as Putty was vital to quickly and easily test both the transmit and receive code that we were using.  
 
Having a few XBee radios laying around your workshop is never a bad idea.  Whether you need to troubleshoot a connection, reduce wire clutter, or need a quick remote control solution, you can't get much easier or quicker than a pair of XBees.
 
Overall, the IEEE Graduate Robotics Team at the University of Tennessee, Knoxville finds the products offered by Parallax essential to the success of our team, and we definitely recommend Parallax products to any robotics team or hobbyist who needs reliable and affordable components for their robotics projects.
 
- Keith Young and the IEEE Graduate Robotics Team at the University of Tennessee, Knoxville