Autonomous Robotic Plane By MIT Skips Obstacles Without Using GPS
Robust Robotics Group from MIT has developed an algorithm for autonomous planes. The best part of the discovery is that it allows the plane to avoid crash with obstacles without using GPS service. To test effectiveness of the algorithm, developers constructed their own plane which contains <span>on-board sensors (accelerometers and gyroscopes), a laser, a basic Intel Atom processor and weighs around 2kg. The plane has exceptionally short and broad wings (with a wingspan of 2 meters ) which allow it to fly at a lower speed.</span>
#-Link-Snipped-#
<span> For initial success of the plane, the team of researchers has provided it with a digital map of the area. The plane has to deduce it's current location on the map and it's orientation. Current location can be identified by using laser and on-board inertial sensors. For orientation (which includes - extent of tilt in a direction, velocity, acceleration), the plane has to perform various calculations. To simplify the calculation process, plane is provided with two different state-estimation algorithms : Particle filter and Kalman filter. Particle filter is time consuming whereas Kalman filter is applicable only in specific conditions. Thus, particle filter is used only in the parts which require it. After computation, the result is again translated into Kalman Filter Language. Developers tested the plane's flight in a parking garage where it covered a distance of 3 miles at a speed of 22 miles per hour, without bumping into anything around.</span>
The flight was really clean, check it by yourself :
<iframe src="https://www.youtube.com/embed/kYs215TgI7c" frameborder="0" width="640" height="360"></iframe>
Via: #-Link-Snipped-#Â | Image Credit: <a href="https://www.dailymail.co.uk/sciencetech/article-2187426/MITs-robot-plane-fly-avoiding-obstacles.html" target="_blank" rel="nofollow noopener noreferrer">MIT's robot plane that can fly itself while avoiding obstacles | Daily Mail Online</a>
#-Link-Snipped-#
<span> For initial success of the plane, the team of researchers has provided it with a digital map of the area. The plane has to deduce it's current location on the map and it's orientation. Current location can be identified by using laser and on-board inertial sensors. For orientation (which includes - extent of tilt in a direction, velocity, acceleration), the plane has to perform various calculations. To simplify the calculation process, plane is provided with two different state-estimation algorithms : Particle filter and Kalman filter. Particle filter is time consuming whereas Kalman filter is applicable only in specific conditions. Thus, particle filter is used only in the parts which require it. After computation, the result is again translated into Kalman Filter Language. Developers tested the plane's flight in a parking garage where it covered a distance of 3 miles at a speed of 22 miles per hour, without bumping into anything around.</span>
The flight was really clean, check it by yourself :
<iframe src="https://www.youtube.com/embed/kYs215TgI7c" frameborder="0" width="640" height="360"></iframe>
Via: #-Link-Snipped-#Â | Image Credit: <a href="https://www.dailymail.co.uk/sciencetech/article-2187426/MITs-robot-plane-fly-avoiding-obstacles.html" target="_blank" rel="nofollow noopener noreferrer">MIT's robot plane that can fly itself while avoiding obstacles | Daily Mail Online</a>
0