We have made great strides when it comes to robotics. But where we've come at a standstill is the lack of support to the robots when it comes to finding the location.


However, Computer Vision has found a solution for this as well. Simultaneous Localization and Mapping are here for robots guiding them every step of the way, just like a GPS.

While GPS does serve as a good mapping system, certain constraints limit its reach. For example, indoors constrain their reach and outdoors have various barriers, which, if the robot hits, can endanger their safety.

And thus, our safety jacket is Simultaneous Localization and Mapping, better known as SLAM that helps it find locations and map their journeys.


As robots can have large memory banks, they keep on mapping their location with the help of SLAM technology. So, recording its journeys, it charts maps. This is very helpful when the robot has to chart a similar course in the future.

Further, with GPS, the certainty with regards to the robot's position is not a guarantee. But SLAM helps determine position. It uses the multi-leveled alignment of sensor data to do so, in the same manner, it creates a map.

Now, while this alignment seems pretty easy, it is not. The alignment of sensor data as a process has many levels. This multi-faceted process requires the application of various algorithms. And for that, we need supreme computer vision and supreme processors found in GPUs.


When posed with a problem, SLAM (Simultaneous Localization and Mapping) solves it. The solution is what helps robots and other robotic units like drones and wheeled robots, etc. find its way outside or within a particular space. It comes in handy when the robot cannot make use of GPS or a built-in map or any other references.

It calculates and determines the way forward concerning the robot's position and orientation concerning various objects in proximity.


It uses sensors for this purpose. The different sensors by way of cameras (that use LIDAR and accelerator measurer and an inertial measurement unit) collect data. This consolidated data is then broken down to create maps.

Sensors have helped increase the degree of accuracy and sturdiness in the robot. It prepares the robot even in adverse conditions.


The cameras take 90 images in a second. It does not just end here. Furthermore, the cameras also click 20 LIDAR images within a second. This gives a precise and accurate account of the nearby surroundings.

These images are used to access data points to determine the location relative to the camera and then plot the map accordingly.

Furthermore, these calculations require fast processing that is available only in GPUs. Near about 20-100 calculations take place within the time frame of a second.

To conclude, it collects data by assessing spatial proximity and then uses algorithms to crack these juxtapositions. Finally, the robot creates a map.


Source by Shalini M