The 43 proposals presented at the Workshop included many designs for airplanes capable of flying in Mars' wispy air (since drones already exist capable of flying in the comparably thin air 35 km above Earth's surface), but surprisingly few balloons.

In the end, only one of the 10 finalists was an aerial vehicle - "Kitty Hawk", from Wendy Calvin of the University of Nevada, who had earlier proposed it as a Discovery mission.

Kitty Hawk consists of four unpowered little gliders, dropped separately to fly over appropriate locations around the great network of valleys that lead into Mars' gigantic 4000-km long Valles Marineris (Marineris Valley). Previously, Dan Goldin had ordered initiation of a project to fly one little actively powered airplane over this region in 2003 to celebrate the first century of powered flight, but this project was quickly dropped due to cost.

Designing a powered airplane for Mars is a lot harder than designing a glider - especially since an engine powerful enough to drive a propeller-driven craft through such thin air must take up a great deal of the plane's total weight, so that only about 10 percent of its mass could consist of scientific instruments.

By contrast, half the total weight of a glider could consist of instruments - and so several of them would weigh no more than a single powered plane with the same payload of instruments, allowing them to cover a total flight distance across the surface that compares pretty well with the flight distance from a single long-range powered plane.

Each of the four Kitty Hawk gliders would travel about 120 km over a period of 10 minutes before crashing. They would be dropped one at a time from an orbiting bus, which would record the high-speed data transmissions from each one and play it back to Earth later.

The Marineris Valley and its tributaries are especially suitable for such aerial reconnaissance because of the very dramatic layering that can be seen in their walls, revealed particularly clearly by MGS' high-resolution photos - both thick layers apparently laid down by successive lava flows in Mars' early days and running the complete height of the canyon's' walls (up to 8 km high), and narrower layers visible on the slopes' lower flanks and running across the flatter canyon floors, piling up again into isolated mesas scattered all over the canyons and towering up to 5 km high themselves.

These latter layers seem to be made of softer, more easily erodable rock, which is probably sedimentary - but how was it laid down? Was it made out of wind-blown dust or perhaps deposited by various episodes of volcanic eruption, before being slowly cemented together by Mars' slow weathering processes?

Or could these be layers of water-borne sediment deposited on the floors of the canyons during Mars' more clement days, either by intermittent torrents of water gushing at high speed through the canyons, or gradually during periods when the canyons may actually have been deep, water-filled lakes?

Clearly, if they're water-deposited, these layers will be especially promising hunting grounds for ancient Martian fossils.

But to understand them and the general geological history of the region, we need both higher-resolution photography of the layering (some of it only a meter or less thick, judging from MGS) and the small channels carved through it in places, and high-resolution IR spectra of the minerals in the individual layers.

Each of the Kitty Hawk gliders would be dropped at a separate location outside one of the canyons, inside a small heat shield, with its wings and tail folded over its main body. At about 15 km, it would unfold - a process already tested on Earth - into a glider with a 2-meter wingspan.

Then navigating by gyros, the glider would angle down to about 5 km, cruise over the canyon's lip, and glide down at an angle following the slope of the canyon wall (which is never more than about 30 degrees), and perhaps making occasional pre-programmed turns to view especially interesting areas.

By the time it reached the canyon's floor, it might be only 2 km up, and would then simply glide across the floor and lower mesas, sending back images to its orbiter until it crashed.

Its instruments would consist of wide and narrow-angle multispectral cameras - with the latter's photos showing details of the canyon layering as small as 10 cm - and a near-IR spectrometer mapping the layers' minerals with a resolution of only 3 meters.

It would have time to transmit only 20 photos, but this would be enough to provide vastly improved understanding of the geology of these great formations - and to judge their value as landing sites for future missions.

One of them would probably be dropped over part of the Candor Canyon where MGS' own IR spectrometer has located what seems to be a smaller deposit of the same coarse-grained hematite which covers a great part of the Simus Meridiani plain - and which may very well have crystallized at the bottom of a lake of standing water.