Juan Pablo is a 3rd year PhD Student in Robotics at Carnegie Mellon University. He received a B.S. in Electrical Engineering in 1996 from Universidad Javeriana in Bogota, Colombia and a M.Sc. in Electrical Engineering from Ohio State University. His areas of interest range from computer vision and image processing to path planning and multi-robot coordination. He has been lecturer in Digital Signal Processing, Electronics and Electronic Design in Colombia. At Ohio State helped develop computer vision algorithms to drive the university's autonomous vehicles during the ITS America Demo '99. He then joined General Dynamics Robotics Systems where he helped develop the Basic Unexploded Ordnance Gathering System (BUGS), a team of small robots for ordnance clearing and disposal. As part of the BUGS project he developed stereo vision, path planning, and autonomous command and control algorithms that allowed the robots to safely navigate in outdoor environments while performing their task. He is currently working with Tony Stentz in applications path planning with uncertainty in position as part of the Collaborative Technology Alliance, and as a Technical Advisor for the Red Team.

I believe we have a duty with those who don't have what we do. I think there is a growing disconnection between high tech and low income, with very few people looking at ways of using technology to improve lives in developing countries.

In spite of recent advances in the provision of accurate environmental data over the globe from satellite data and other sources, there is still a lack of comparable high resolution soil data. Detailed soil information exists for only a small fraction of the globe. The last major effort to produce global coverage of soil maps dates back to 1974, with the publication of the FAO soil map of the world, hence soil data now lags seriously behind information of comparable environmental attributes.

Soil maps lack not just precision, but accuracy. Conventional soil maps use concepts that lack at least 50 years. They are predominantly qualitative, and depend on poorly specified predictive models that are not updatable.

The International Center for Tropical Agriculture (CIAT) is interested in improving the quality of soil maps for the tropics, therefore improving its ability to:

• visualize catchment hydrology at a scale amenable to community-based management
• help target soil-sensitive crops confidently within new areas
• help explain complex patterns of changing land use that underwrite landscape resilience.

The proposed work will explore ways to use the knowledge of the scientists at CIAT and existing GIS data to create soil models for the tropics using artificial intelligence, machine learning and/or data mining techniques.