The list of accepted papers for ACM GIS 2008 (to be held this November in Irvine) is out, and there's a lot there that looks of likely interest to computational geometers.

My own contribution with Mike Goodrich is Studying (Non-Planar) Road Networks Through an Algorithmic Lens. The technical content of the paper concerns using separator decompositions and various related tricks to speed up algorithms for network optimizations such as shortest paths (allowing arbitrary edge weights, as needed to model user preferences for trading off speed, distance, toll cost, scenicness, etc.) but the main point of the paper is to point out that real-world road networks are far from the planar graphs they are often erroneously assumed to be, and to attempt to model them more accurately as subgraphs of low-ply disk intersection systems building on Shang-Hua Teng's neighborhood graph research.