This page is no longer being updated and remains online for informational and historical purposes only. The information is accurate as of the last page update.
For questions about page contents, contact the Communications Division.
Serdar Benderli ’09 (Antalya, Turkey) is pursuing a B.S. in electrical and computer engineering and an A.B. in economics and business. This year he is working on an independent study under the guidance of John Greco, professor of electrical and computer engineering. Benderli is working to develop a micro-mouse robot that can find the shortest path through an unknown maze. The following is a first-hand account of Benderli’s experiences with the project.
I am working on an independent study with Professor Greco of the ECE department on designing a robot to solve an unknown maze. The goal of the robot is to find the shortest path to the center of the maze and get back to the starting point in the shortest time possible. Only the size of the maze (i.e. number of cells it contains and the dimensions of each cell) are known prior to the exploration. Since we are planning to enter Micro-Mouse competitions held by the IEEE (Institute of Electrical and Electronics Engineers) sometime next spring, there are some further constrictions on the robot that we have to take into consideration.
I started working on the robot last summer in my free time to get myself acquainted with the project. After the semester started, I constantly met with Professor Greco to discuss the current situation, to come up with solutions, and ways to improve the design. The strategy is usually to tackle problems one at a time, because otherwise it gets encumbering. You’d be surprised how many problems you can solve when you break them apart into simpler, smaller parts.
The robot essentially has two motors and a bunch of sensors to find its way through the maze. These motors and sensors are all driven by a microcontroller, which holds the algorithm to solve the maze. As the robot moves through the cells, it senses the walls around it and decides on its next step until it finds the center.
The main problem is keeping the robot straight and on target. This is called odometry and it is a problem because any deviation from straight could make the whole search fail. These deviations could be caused by a number of reasons, such as offsets during left-right-back turns, one wheel turning more than the other, slips due to dust on the floor, etc. Since there are no indicators or signs on the cells themselves, it is all up to the robot to keep itself straight and on track.
To achieve this, we’re using a feedback system which tells the microcontroller how much each wheel has turned while moving. Thus, the microcontroller can decide based on these values whether or not both wheels turned the same amount. If there’s any error, say the right wheel moved a bit more than the left, the microcontroller would advance the left wheel at the same amount and get back on track.
To give an example of technical problems that come up, we just realized that the microcontroller we have is not enough to solve the maze, drive the motors, and keep the robot straight at the same time. Thus, we decided to divide things up and use a second microcontroller to do the moving and the odometry. Since we chose a microcontroller that uses a different architecture due to some technical reasons and budget constraints (the IEEE demands the robot costs less than $500 altogether), communication between the two will be a hard issue, not to mention learning how to program in a different language and very different architecture.
As you can see, the project is full of challenges. But, that’s what makes it even more interesting and exciting. To create problems and then try to solve them is what engineers do, and the problems with this project are constantly in the back of my head, bugging me at every opportunity, and I really enjoy that. Adding on to that is the remarkable opportunity to work with one-on-one and learn from a professor as experienced and as fun as Professor Greco. I am not only gaining experience in robotics, but also in embedded systems development, an area I’m very interested in working after college. These are extremely valuable assets for an electrical and computer engineer and I feel very lucky to be involved in such a project.
When we enter the competition, I hope to kick some robot butt with EftiBot, named after my girlfriend whose constant support made many long nights of work endurable.