|   ||   |
A Micro Air Vehicle to Fly in Caves, Tunnels, and Forests
Recently, there is a need to acquire intelligence in hostile or dangerous environments such as caves, forests, or urban areas. Rather than risking human life, backpackable, bird-sized aircraft, equipped with a wireless camera, can be rapidly deployed to gather reconnaissance in such environments. However, they first must be designed to fly in tight, cluttered terrain. This research discusses an additional flight modality for a fixed-wing aircraft, enabling it to supplement existing endurance superiority with hovering capabilities. An inertial measurement sensor and an onboard processing and control unit, used to achieve autonomous hovering, are also described. This is, to the best of our knowledge, the first documented success of hovering a fixed-wing Micro Air Vehicle autonomously.
Visual Servoing on a Line Using PD Control
A wireless camera, mounted on the gondola of a blimp, transmits images back to a ground-based PC. The PC processes the images to find the line's orientation, implements PD control, and then transmits a signal back to the blimp's yawing motor to visually servo on the 15-foot line.
Adding Intelligence to ER1 Robot
Artificial Intelligence algorithms, namely A* Search and Potential Field Navigation, are implemented on Evolution Robotics' ER1 Personal Robot System. Given a 10 x 10 foot state space populated with several obstacles, the ER1 is able to successfully reach a goal state without colliding into any obstacles.
Closed Quarters Aerial Robotics (CQAR)
Urban environments are time consuming, labor intensive and possibly dangerous to safe guard. Accomplishing tasks like bomb detection, search-and-rescue and reconnaissance with aerial robots could save resources. CQAR is a Closed Quarter Aerial Robot which is capable of flying in and around buildings. The prototype was analytically designed to fly safely and slowly.
Optic Flow Based Autonomous Tasks
Insects make heavy use of vision, especially optic flow, for perceiving the environment. Optic flow is the apparent visual motion of the surroundings when traveling through the environment. With optic flow sensors, algorithms can be developed that will allow UAVs to navigate through near Earth environments by mimicking the natural behaviors of insects.
MAV Optic Flow Simulation
Evaluation of control algorithms through 6 DOF flight tests is time and cost intensive. A simulation which mimics near-Earth environments is a more efficient method for fine tuning control strategems. Ultimately, this simulator would allow parameters (i.e. proportional constants) to be adjusted for different environmental conditions (e.g. lighting levels).
- Download MAV tunnel simulation
Application of Neural Nets to Optic Flow Sensors
Optic flow sensors are sensitive to changing lighting conditions among other things. For example, moving the sensor past the same scene (i.e. texture and contrast are the same) in environments with different lighting conditions (e.g. cave and a gymnasium) will yield different magnitudes of optic flow output. However, applying neural networks to characterize the sensor output will allow similar results with varying environmental conditions.
Civilian Medical Response Center (CiMeRC)
Extensive interviews with first response teams and site commanders reveal that verbal communication between parties in dynamic environments, like disaster areas, are often misinterpreted. Using a two-way radio to convey information like situational awareness, tasking commands or resource availability can often be misunderstood or misinterpreted, because the voice-based command and control always has the possibility that questions and answers will be vague or ambiguous. A picture's "thousand word worth" can be used to augment the communication efforts. In my previous research, I helped prototype an aerial video acquisition and imagery distribution system where raw video images are processed and can be wirelessly downloaded by first responders equipped with handheld devices like cell phones or Palm Pilots. Aerial images present useful information that might not be seen from someone on the ground such as the extent of damage, structural integrity of buildings and bridges, and ingress and egress routes to the site and nearby hospitals, respectively.
Kite and Teleoperated Vision System
In times of disaster acquiring aerial images is challenging. Runways may be crippled thus denying conventional aircraft in the area from taking off. Also the time required to schedule a satellite fly-by may delay first response efforts. Man backpackable aerial robots can be carried close to the disaster site and flown to capture aerial images. Mechatronics, intelligent sensing, and mechanism synthesis is integrated in a teleoperable kite-mounted camera. Rapidly deployable, transportable by foot, easy to fly and affordable, our system can quickly acquire, process and distribute aerial images. Image mosaicing, edge detection, 3D reconstruction and geo-referencing acquired images is also discussed.
- View other system applications