Cookies   I display ads to cover the expenses. See the privacy policy for more information. You can keep or reject the ads.

Video thumbnail
Here at the Computer Science and Artificial Intelligence
Laboratory at MIT, we've developed a nonportable mapping
system which enables exploration in GPS [? in all these ?]
buildings and indoor areas, allowing the user
to build maps in real time as they explore their environment.
The device worn by the user contains onboard processing
in a backpack, a Kinect depth sensor, an inertial sensor,
and a ranging LIDAR, or laser rangefinder.
As the user explores, his motion is
determined using incremental LIDAR scan matching.
The LIDAR sweeps a laser beam around in a 270 degree arc,
and measures the time it takes for the light pulses to return.
Reprojecting the LIDAR scans produces
this continually expanding map.
However, motion drift will gradually
cause errors in the map.
Errors can be rejected when the user returns to a location that
has been previously observed.
In addition, it is important that these scans be
corrected for the user's gauge.
This is done using the inertial sensor.
The entire process is real time.
All the necessary computation is carried out
on the explorer's backpack.
Meanwhile, the camera system collects
snapshots which you can see in the bottom right.
It's these images that uses to detect
a previously visited location.
During larger excursions, significant drift can occur.
These can visibly corrupt the map.
For example, obscuring doorways or explore unexplored areas.
When a previously visited location is determined,
map smoothing can resolve this inconsistency.
Using a clicker, the user can inject tags into the map,
labeling important or interesting locations.
In the future, we hope to annotate the map with higher
level information, such as spoken directions
or detected signage.
What you don't see in this video is that the device also
supports multi-floor mapping by detecting operation
in staircases and elevators, using the inertial sensor
and a barometer.
These maps can be transmitted wirelessly in real time
back to a remote bay station.
The goal of this project is to enable situational awareness
by the user or an external commander
in search and rescue operations.