One area of multi-rotor UAV research that has been barely scratched is ground locomotion. The definition of ground locomotion is the use of systems built into the flight vehicle to move on the ground, likely without using the flight propulsive systems. Basically, this is what you get when you combine a UAV with a UGV. The following points illustrate the reasoning and need for such a system:
1) Electric Multi-rotor UAVs are only limited in flight duration by the battery capacity they carry. So, unless they exhibit some type of energy-harvesting behavior (mainly solar- or laser-powered flight), multi-rotor UAVs must land, usually sooner than later as compared to a fixed-wing UAV.
2) Although Multi-rotor UAVs are well suited to human-operated or human-assisted missions such as aerial photography, bridge inspection, and so on, there is little reason against, and many possibilities for, development of a completely autonomous system that does not require human supervision. This includes perimeter monitoring, migratory vehicles, scientific sampling, and other tasks. To create such a system, the vehicle must be able to autonomously recharge or exchange its batteries.
3) At present, most autonomous multi-rotor UAV research takes place in an indoor laboratory where researchers directly handle and maintain the test vehicle. Unless the research focus is the development of a recharging plate or platform for the test vehicle to land on to recharge (such as has been done at ETH Zurich), the autonomous system is only self-reliant in matters of flight and aerial maneuvering. Rendered another way: the machine must undergo the equivalent of a major surgical operation (preformed as the researcher removes and replaces the battery pack) in order to remain an active flying machine.
4) If the research topic is autonomous recharging, multicopters work great in an environment where a well-tuned motion-capture system and developed charging interface exists. However, as soon as autonomous multicopters venture outside of the lab into the outdoors, the well-defined and controlled environment disappears. Multi-rotors can exist in this type of environment just fine so long as attitude state information is known, but position information is somewhat reduced: The cues from a GPS and other navigation source can have a mean error of up to 2 meters, far to large for a precision landing on a charging pad.
5) A motion-capture system could be installed at the critical area of a landing pad or recharging dock/station to forward accurate position and state data to the vehicle. A multi-rotor could use GPS to arrive in the vicinity of the target area, to have the motion-capture system pickup the reflective patterns on the UAV, thus accomplishing the same precise interface as in the lab. Conversely, the UAV can have the vision-systems/devices necessary to use cues from defined targets on the landing dock.
6) For both versions of the solution posited in Point 5, there is still a large degree of complexity. Either the UAV must communicate with the ground station motion-capture systems, or the UAV must be able to identify specific marks on the charging platform. This ignores the other complex tasks involved in the scenario such as identifying if a station is already occupied by another quad/multi-rotor, and the possibility for unusual lighting and optical difficulties that might make vision-based landing difficult.
7) The alternate solution introduced here: At the cost of about 40 to 60 grams, a pair of continuous-rotation servos with wheels can be mounted to the underside of the quadrotor. Instead of negotiating with a motion-capture data link or processing data from a computer vision system to make a perfect landing on the charging pad, the entire flight vehicle can make a landing using GPS to within a flat 3 to 5 meter square target area. On the ground, the vehicle can drive directly to a empty charging dock using methods that have long existed for UGVs, such as line-following or lighthouse type arrangements. The control problem is simplified in that no motion capture system needs to exist, and the design of the landing area can be simplified or done away so long as the vehicle can drive in the landing area unhindered and locate a charger readily.
8) The utility of a ground locomotion system is much more useful to a multicopter than for just charging: Positioning on the top of a building for a better camera shot, handling the logistics of entering and exiting structures and areas that subdue flight devices, and accomplishing movement in general without using the flight devices are just a few examples.
Bibliography and Related Work:
VTOL Air/Ground Design: http://diydrones.com/profiles/blogs/transformer-bot-turns-from
ETH Zurich Recharging Pads: http://youtu.be/pcgvWhu8Arc (Seen from beginning and 0:25)
Attempts to have a multi-copter land on a UGV: https://www.youtube.com/watch?v=XpUdW_U2KJ8
Ground-based recharging: Seen in all commercial household cleaning robots such as the iRobot and Neato.
Comments
There are inductive style contactless proximity charging technologies available now commercially, such as the Qi standard. It might be possible to use a larger Qi pad and autonegotiate to higher power transfers than those used for charging cellphones, since there is some flexibility in the standard, and you would prefer not to have to wait a few hours to be able to fly 10 minutes again. Couple this with something like a downward facing camera on the UAS that doubles as an optical flow sensor, and place ranging markers on the pad with a QRcode printed in the center. If the camera can't spot the QRcode, assume the pad is denied by debris or another vehicle.
Having low powered (for various meanings of "low") ground locomotion for a UAS is interesting, if only as a kind of range extending, though that rapidly approaches the line between a car that can fly and an aircraft that can drive. You almost start to emulate the DARPA Transformer flying car project. Which kinda brings back the issue of whether such a setup would be better served by a paired UGV/UAS that can dock/link to each other to operate as a single vehicle, but can also operate independently or as a coordinated team. Slightly solves the landing issue since you would have an active/active setup.