Preliminary project information & guidelines

• The project should be an in depth exploration and implementation of some topic in robotic manipulation or mobile robotics. The size of the project should be appropriate for its two month duration; you'll also note that there are no assignments due in the second half of the class so you can focus more on the project.

• I strongly encourage teams of 3 +/- 1. Particularly in CSCI 4961/6961, I also encourage mixed teams of students from different majors.

• You can consider two types of projects: hardware and software projects. Generally speaking, both types of projects should implement or develop some algorithm(s), with hardware projects ultimately executing these algorithms on physical hardware; software projects might simulate algorithm execution.

• Suitability and scope of projects are to be negotiated with Prof. Huang.

• Projects should have some "scientific" interest. I expect that there should be at least a little bit of a literature search to uncover relevant background information in conference or journal papers.

• Projects should have more "science" than "engineering". For example, hardware projects should not focus on building some sort of robot; the focus should instead be on developing algorithms for some task/system.

Project activities

• Written project proposal. We can "renegotiate" this as necessary as the project progresses.

• Phase 1 of your project is due at the end of March. The basic idea here is to have something due both to keep you on track and so that your entire project grade is not assigned at the very end. The exact details of this will be negotiated for each project. Further guidelines will be forthcoming.

• Class presentation at the end of the semester: each team will give a brief description and summary of their project.

• Demonstration: to be arranged with Prof. Huang and/or Richard Fauconier at the end of the semester.

• Written final report

Project suggestions

Where to look for project ideas

• IEEE International Conference on Robotics and Automation (conference proceedings, TJ 211 I4x)
• International Journal of Robotics Research (journal, 629.892 I54)
• IEEE Transactions on Robotics and Automation (journal, 629.892 I35)
• IEEE/RSJ International Conference on Robot and Systems (conference proceedings, TJ210.3 .I443x)
• Workshop on the algorithmic foundations of robotics (workshop proceedings, TJ211 .W665 1994, TJ211.4 .W64 1996, TJ210.3 .W664 1998) I think I'll put these on reserve...
• Robot Motion Planning by Jean Claude Latombe. On reserve for this class.

The electronic database (which covers robotics) is the INSPEC database. From the RensSearch home page, select "Engineering" under the "Databases that cover:" pull down list. You will get a page that has a link to the INSPEC database (among others which aren't really relevant). Note that the INSPEC database covers publications that the library doesn't have.

Hardware project suggestions

• SONAR mapping with mobile robots: use the SONAR sensor to create a map of the lab and use it for path planning, navigation, and localization. There are several different approaches to this problem that different groups can try, from fairly simple "evidence grids" to more sophisticated (and more robust) bayesian approaches.

• Mobile robot coverage: path planning and control for a mobile robot that must pass some sensor or actuator (think of search and rescue, or mowing the lawn as applications) over every point in an area.

• Yoyo project: create a large yoyo and tie the string to a motor. The objective is to get the yoyo to go up and down with a periodic motion.

• Cup in ball (kendama): using a 1 degree of freedom arm, place a cup at one end and tie a string with a ball attached at the end. The objective is to plan an arm motion that swings the ball up so that it lands in the cup.

• "Pachinko" parts feeding: Prof. Akella has worked with an array of pneumatically controlled "pins". You can manipulate planar parts on this array (think of the array being at a slat so that parts would slide down if they are not being "held" by the pins). The objective would be to create plans that manipulate parts by pulling in these pins at appropriate times.

Software projects

• Nonholonomic motion planning: implement a motion planner for nonholonomic motion platforms such as a car-like robot or a differential-drive robot. You can also extend this to add "trailers" to these vehicles. There are several different approaches to this problem.

• (Multiple) mobile robot coverage: path planning and control for one or more mobile robots that must pass some sensors or actuators (think of search and rescue, or mowing the lawn as applications) over every point in an area.

• more to come...