Your ad can be shown here!
 
  Help
AI links
  :: main hierarchy ::
• Agent Software
• Artificial Life
• Cellular Automata
• Cognitive Science
• Companies
• Computer Vision
• Data Mining
• Decision Support
• Expert Systems
• Fuzzy Logic
• Game AI
• General Resources
• Genetic Algorithms
• Handwriting Recognition
• Information Retrieval
• Intelligent Agents
• Intelligent Interfaces
• Knowledge Management
• LISP
• Machine Learning
• Mobile Agents
• Nanotechnology
• Natural Language Understanding
• Neural Networks
• Online Books
• Philosophy
• Programming
• Prolog
• Robot Builders
• Robotics
• Speech Recognition
• Virtual Pets
• Web Agents
 
aboutAI.net Weekly Features
    Let's Fly Again
Let's Fly Again
What's new in the field of aerial robotics, featuring results from the 2001 International Aerial Robotics Competition and details on home-built UAVs that can fly to the edge of space.
  Related Resources
• Past issues of weekly features
• Micro-Air Robots
• Learning to Fly
 
 Elsewhere on the Web
• Remote Control Helicopter Links
• Team Papers for the past IARCs
 
 

"The 2001 International Aerial Robotics Competition was held on the weekend of July 21 at Webster Field in the state of Maryland. Seven teams attended during this first year of the new mission, and the team from the Georgia Institute of Technology successfully demonstrated the Level 1 behavior on its first attempt." This news brief instantly drew my attention, as I'm receiving numerous comments and questions related to Unmanned Aerial Vehicles (UAVs) and other air robots. An earlier article on micro-air robots is a good starting point for further research on this topic, but this time we'll learn more about new products and research projects.

Back to the Aerial Robotics Competition, seven teams showed up at the competition this year, including Simon Fraser, Rose Hullman Institute, North Carolina State, Southern Polytechnic University, Purdue University, University of Waterloo, and the winning Georgia Tech. Waterloo only had a presentation, and all other teams except Georgia Tech brought helicopters. The main task involved a demonstration of fully autonomous flight over a large area in an attempt to perform one of three different missions:

  • Hostage rescue: robot launched from a submarine must locate hostages
  • Nuclear disaster - robot transports sensors to the damaged nuclear reactor complex
  • Biological emergency - robot returns data from a quarantine area
Common to all three mission examples is the ability to fly to a specified location from a distance of 3 kilometers and identify a particular structure. Once the structure has been identified, a sensor probe must be sent into the structure to perform reconnaissance of a particular type. The Level 1 behavior I mentioned above is defined as an "autonomous flight over a distance of 3 km beginning at a designated starting point and terminating in an autonomous hover or orbit about a designated final way point, with up to four other way points visited along the path." Levels 2 - 4 require progressively advanced behaviors, including sophisticated computer vision and robot navigation techniques. The first team that exhibits the Level 4 behavior will receive a large cash prize that increases by $10,000 per year until won.

Unfortunately, Georgia Tech was the only team to accomplish any kind of autonomous flight at the competition this year. Even the winning team wasn't aware of all the requirements for Level 1. Technically speaking, probably the most impressive piece of hardware aboard their vehicle was MicroPilot's MP2000 autopilot. This low cost ($5,000) autopilot is small enough to fit into an RC trainer, with capabilities that include airspeed hold, altitude hold, turn coordination, GPS navigation, autonomous takeoff and landing, etc. Data logging and manual overrides are also supported. All feedback loop gains and flight parameters are user programmable, and PC based mission simulation software enables user to gather more experience without destroying several UAVs. Less advanced configurations are also available, starting at $3500 for a complete, ready-to-fly UAV or $1500 for MP1100 Autopilot with GPS receiver, GPS antenna and connector kit.

Some of the unrelated UAV projects are equally fascinating. Several groups are really pushing the limits by building UAVs that should fly to the edge of space. T.J. Bordelon and his team, FreeSpace, are working on ultra-high altitude balloons and robotic aircrafts. Their RadioFlyer-1 is a balloon platform that reached halfway into space (94,000 feet) and took some perfect photos and telemetry data. Shadow-1, on the other hand, is an off-the-shelf model airplane that can "fly by itself". The goal of the shadow projects is to replace the balloons for a more controlled trip into the stratosphere. When this project is complete, T.J. is considering offering a complete UAV developers kit, along with a CD of stuff to help you with your own project. If you're interested in this sort of package, please give him feedback on what you'd like to see. His home page also lists other balloon groups and UAV-related links.

Some of the advanced research projects mentioned in the previous article recently started to produce first practical results. The Center for Intelligent Mechatronics at Vanderbilt University is developing mesoscale robotic insects that are capable of efficient terrestrial locomotion. Stefan Marti of the MIT Media Laboratory still works on Free Flying Micro Platform (FFMP) and Papa-TV-Bot, autonomously hovering mobots (mobile robots). His FFMP bookmarks collection, although uncommented and a bit outdated, is one of the best resources for a starter. Just point your favorite Web spider to this list, and you'll get hundreds of pages filled with interesting info. If you are interested in commercially available UAVs, NASA GSFC/Wallops Flight Facility maintains the UAV Characteristics Database with more than 40 detailed descriptions of each model. Another similar resource is UAV Forum, filled with info on all sorts of commercial air robots.

Talk about AI-related topics at Artificial Intelligence Bulletin Board.

Want more timely information and resources related to AI? Subscribe to our FREE newsletter!

Got some specific AI related questions or need consulting services? Contact your webmaster, Denis Susac.

Talk with people who share your interests...

Previous Features

S p o n s o r e d    b y...

Buy the ER1 Robot!