SURF 2008

This page is intended for students interested in working on SURF projects in the Summer of 2008. It contains a list of project areas where I will be supervising projects this year. Students interested in writing proposals for SURF projects should contact me via e-mail to discuss what areas they are interested in and talk through possible SURF proposal topics. All applications should go through the normal SURF application process, described at www.surf.caltech.edu.

Project Areas

Autonomous Vehicles

Alice

Caltech participated in the 2007 DARPA Urban Challenge. We made it to the semi-finals, but failed to make it to the race this year. A detailed description of our system, including an analysis of the limitations of our system is available in the final report for our project [1]. The goal for this summer is to extend the capabilities of Alice so that it can drive on urban streets without accurate prior information about the location of the roads. This will require advances in a number of areas:

• Detection and interpretation of road markings - we need to improve our line and curb detection algorithms and fuse this into a map of the road in front of us. Complicating factors include the presence of shadows, vehicles (moving or parked) and variations on road markings between different types of roads.
• Dynamic planning algorithms - without prior information on road locations, we will have to do dynamical plan paths that depend on the real-time road information (which may be noisy)
• Higher level logic for intersection handling and merging - our current logic handles intersections reasonably well, but needs improvement in handling merging into traffic (the test that we failed at the NQE)
• Modeling and simulation environments - in order to effectively test our algorithms, we need to develop some good modeling infrastructure that allows different levels of simulations to be combined and integrated in an effective manner. Especially important is the inclusion of sensory-based simulations that can feed representative raw data into low level perceptors based on a simulated environment

This project can support 4-6 students who will work together as a team. There are currently 2 students who have expressed interest in the general area of the project.

References:

1. J. W. Burdick et al., "Sensing, Navigation and Reasoning Technologies for the DARPA Urban Challenge. DARPA Final Report, 2007.
2. L. B. Cremean et al., "Alice: An Information-Rich Autonomous Vehicle for High-Speed Desert Navigation". Journal of Field Robotics, 2006.

Titan

The Jet Propulsion Laboratory (JPL) is working on a mission to Titan (one of Saturn's moon) that will involve flying a balloon in the atmosphere of the moon to take scientific readings of the ground. This project involves several faculty from Aeronautics, Mechanical Engineering and Control and Dynamical Systems. The projects in my group will focus on evaluating autonomous controls technology that can be used on a prototyped (to be developed as part of CS/EE/ME 75 in 2008-09). Specific projects include:

• Controllability analysis for vehicle configurations - we expect to explore a number of different vehicle configurations (locations of thrusters, mechanisms for controlling orientation, etc). An important evaluation criterion will be the controllability of the configuration: how well can the vehicle maneuver in terms of satisfying the science objectives of the system.
• Receding horizon control using observed/estimated atmospheric currents - the local (100 km scale) atmospheric environment around the vehicle must be used to determine how to achieve mission objectives, but this environment may not be directly measurable. This project will explore the use of receding horizon control techniques for trajectory planning and control in the face of atmospheric uncertainty.

This project can support 1-2 SURF students. Students should have taken CDS 110ab or equivalent.

References

1. Case Study: Titan Surface and Subsurface In Situ System

Synthetic Biology

Molecular Programming

A molecular program is a collection of molecules that may perform a computation, fabricate an object, or control a system of molecular sensors and actuators. The best examples are the biomolecular programs of life---from the low-level operating system controlling cell metabolism to the high-level code for development, the process by which a single cell becomes an entire organism. Life's biomolecular programs are intrinsically parallel, distributed, asynchronous, error-prone yet amazingly robust processes; this shows that issues central to computer science are intrinsic to molecular programs.

This project will explore the role of feedback in molecular programs and the design of specific feedback circuits using in vitro biochemistry. Possible project areas include the design of PID (or PI or PD) controllers using biopolyners (DNA and RNA). The projects should involve a combination of modeling, experiments and analysis.

References:

1. E. Franco, P.-O. Forsberg and R. M. Murray, Design, modeling and synthesis of an in vitro transcription rate regulatory circuit. Submitted, 2008 American Control Conference.
2. Georg Seelig, David Soloveichik, David Yu Zhang, Erik Winfree, Enzyme-Free Nucleic Acid Logic Circuits. Science, 2006.

Biological Circuit Design

Additional Information

I can only take on a limited number of students in each project area. Unlike previous years, where we have preselected students to participate in writing a proposal, this year I will use the SURF proposal process to sort out the students that will be accepted into the program. This means that your SURF proposal will be compared against others in the same area and the only top proposals will be selected. To help avoid having a large number of proposals rejected, I will post information on this wiki page as students sign up for project areas, so that you know how many other proposals are being written.

Because there are quite a few students who have expressed interest in writing a proposal, it will not be possible for me to work closely with each of you in selecting a project area. Hence, you should do your homework and read about the project areas before contacting me about the possibility of doing a project. Sending me an e-mail expressing interest in a project area without having read some of the references that I have listed is not a good start! I will use this wiki page to post answers to questions and additional information about the projects, so that everyone has the same access to the information.

As a starting point to writing a proposal, I suggest that you consider putting together a GOTChA Chart. This is a simple planning tool that helps nail down the essence of a proposal and can then be used as an outline for the official SURF application.