You are currently browsing the tag archive for the ‘unmanned vehicle systems’ tag.

There was two interesting pieces of aero/astro related news in this week’s IEEE Spectrum. First, the announcement by Russia of a new manned space program, whose goal is to send humans to Mars. More than the talks about international cooperation mentioned in the article, I think (unfortunately perhaps) it’s only international competition and the development of new space programs in Russia, China, India, etc., that will settle the current relatively sterile debates around Nasa’s own programs, in particular going to Mars or back to the Moon.

And a funny new UAV, developped by a German team. The applications would be the usual ones, surveillance and emergency communications. But what’s more interesting is that there is absolutely no pilot in the loop here controlling the aircraft remotely. So this leaves us with a lot of opportunities to improve autonomy technology and safe integration into the airspace.

Advertisements

Last week the Year 1 program review meeting for the autonomy center of the MAST project took place here at Penn. This project, for which you can find more information on the ARL and MAST websites, is concerned with the development of teams of small mobile robots (air and ground vehicles) performing  various missions for military applications, essentially increasing situational awareness. MAST is a very large project, and is divided into 4 main research areas (or centers) representing different aspects that need to be developed: platform mechanics, microelectronics, processing for autonomous operations, and integration. The goal is to combine the developments of the different centers into one integrated demo by the end of the project.

GRASP is the lead institution for the autonomy center, which also includes participants from the ARL, Georgia Tech, UC Berkeley, UC Merced, University of Mexico, University of Delaware, University of Maryland, MIT, and the University of Sidney in Australia. The researchers span the fields of robotics, control, vision and sensing, and communication systems. A very interesting aspect of the project is that it is bringing these usually separate communities together to work on common problems and combine their knowledge. To take an example, Yasamin Mostofi is cooperating with Bert Tanner to develop motion planning strategies for groups of robots that use realistic models of the wireless communication links between the robots. The wireless communication systems community usually does not consider controlled mobility of the nodes in their models, and typically models the motion as random. The researchers working on motion planning for robotic networks on the other hand have proposed models taking into account communication constraints, but usually these communication models are oversimplified. The popular disc model for example assumes that the robots can communicate with their neighbors if and only if they are within a prespecified radius, but this does not model properly the complicated wireless interference aspects, especially those arising in indoor environments where multipath fading effects are important. They also want to integrated the motion planning component with the sensing task, for which Songhwai Oh is using environmental models based on Gaussian processes.

update 04/11/08: An article describing the HUNT workshop at ASU was published by insciences.org

February has been a hectic month and unfortunately I haven’t been able to update this blog as I hoped. But I have several posts in mind and I might be able to finish some of them in the coming weeks.

I’m just returning from a two-day workshop at ASU for the HUNT project (Heterogeneous Unmanned Networked Teams) that I’m involved with. The link will take you to the workshop program. This project is about the control of teams of unmanned vehicles, and there is a strong focus on designing bio-inspired group behaviors. So this time the first day of the program was reserved for talks from biologists, studying lions, monkeys, fishes, ants, among other species.

I have to say first that I’m impressed by the results that have been possible in robotics thanks to biomimicry. A good example at GRASP is the RHex robot built by the group of Dan Koditschek. But so far, most successes seem related to robot mechanics and hardware technology. For robot AI, although behavior-based robotics is largely inspired by biology (incidentally, Ron Arkin is a PI in HUNT), I haven’t found a good example yet which demonstrates that studying animal groups will help us design autonomous robot teams performing complex missions. Hopefully we’ll see¬† one coming out of this project. You might say, what about swarming, flocking, and the related topics that have drawn so much attention in the past decade? Well, I don’t think this is addressing the high level mission planning problem that UAVs will need to solve, such as data collection tasks. But since this might be a bit controversial, I’m planning to write more on this subject. For now, Magnus Egerstedt pointed us to this video, which shows what we can currently do with a good amount of cheating ;-)