Parrot, Canonical
make a S.L.A.M.dunk to make drones smarter
The world of unmanned aerial
systems has been rapidly growing and uncovering new challenges to their ideal, practical,
and efficient utilization. One of the biggest hurdles in this matter has been
achieving the point where the small drone or UAV has the ability to avoid
obstacles autonomously. The article from the Splash Gear site presents an
innovative piece of technology being introduced into the market later this
year, which would further a small UAV’s abilities in regards to autonomy and
obstacle avoidance through environment 3D mapping.
The device was developed by a
partnership between Parrot, a renowned company in drone development, and
Canonical. It is called the S.L.A.M. dunk; short for “Simultaneous Location and
Mapping”, and it is meant to be mounted horizontally to a drone serving as its
eyes. It is a development kit equipped with sensors and processors including an
NVIDIA Tegra K1 processor, a 1500×1500 fish-eye stereo camera, an Inertial
Measurement Unit (IMU), an ultrasound sensor, a magnetometer, and a barometer.
These enable the drone to receive information and formulate commands to avoid
obstacles in its intended or pre-programmed path.
Canonical brings the software side
of the operation to the S.L.A.M. dunk. Using Ubuntu and Robot Operating System
(ROS) this kit is highly desirable among developers for its flexibility in
experimentation and modifiability. The device also provides mobile computing in
that when connected to a monitor, it enables access to a regular Ubuntu
desktop.
Although there are autonomous drones in the
market already, they are generally very expensive and are hardly modifiable or
open to development. Most affordable drones are not autonomous and are limited
in the operational distance from an operator who remotely sends the commands.
With this device autonomous tasks can be programmed such as mapping, imaging,
or even walking the dog.
Torres, J. (2016, September 9). Parrot, Canonical
make a S.L.A.M.dunk to make drones smarter. Retrieved from Splash Gear:
http://www.slashgear.com/parrot-canonical-make-a-s-l-a-m-dunk-to-make-drones-smarter-09455455/
Miguel,
ReplyDeleteNice setup for the article review. Great technology advancement for UAV's to have split second avoidance of obstacles. Algorithm usage and software advances are so good today. Good job.
Jason
Miguel,
ReplyDeleteI have never heard of this system, now I know what sensor to use on UAV to avoid obstacles. Simultaneous Localization and Mapping (SLAM) problem can be defined as a process where a robot builds a map representing its spatial environment while keeping rack of its position within the built map. Mapping is done online with no prior knowledge of the robot’s location; the built map is subsequently used by the robot for navigation. SLAM is a key component of any truly autonomous robot. The basic SLAM framework involves odometry, landmark prediction, landmark extraction, data association and matching, pose estimation, and map update. These processes are the backbone of every major SLAM method, and are performed in cyclic fashion.
Frank
Miguel,
ReplyDeleteExcellent blog. This is particularly new to me, and I find it to be a very intriguing "sense and avoid" system. I like the concept of all-in-one packaging of sensors, like the stereo camera, IMU, magnetometer, barometer, and ultrasound sensor. I hope that it can be easily adaptable to any UAV, especially with the S.L.A.M. duck development kit from Parrot, which according to dronelife.com, is expected to be available soon at http://developer.parrot.com/ in the Q4 2016.
Miguel,
ReplyDeleteGreat review on the new technology for sensors. It is amazing that the integration of software and sensors is this advanced. The use of this system will make the vehicle easier to control and more autonomous. It will be amazing to see what this technology evolves to in the future.
Tyler
Miguel,
ReplyDeleteI think you really covered this emerging tech well, and as a sports fan, it doesn't have a bad name either. I can see this tech being used in 3D printing for creating scale models for inaccessible cave systems or other architecture applications.