| Summary: | Traditional SLAM (Simultaneous localization and mapping) robots made use of techniques
such as the extended Kalman filter and fastSLAM. The extended Kalman filter is a recursive
technique whereby estimation of the current state is based on measurement from the previous
state. Firstly, the algorithm predicts the state of the robot based on some sensor values.
Subsequently, the actual state of the robot is observed based on other sensor values. Actual
state of robot may be wrong due to errors in sensor. Finally, the algorithm compares the two
values and determines the best location for the state of the robot. The process is repeated for
the subsequence state of the robot [1].
The above short explanation demonstrated the limitation of implementing such algorithms
into low cost SLAM robots as the localization depends heavily on the precision of the sensor.
A precise sensor is needed for accurate estimation. This limits the possibility of implementing
SLAM into embedded systems found commonly in our daily life. The project look into the
possibility of using the raspberry pi 3 embedded system board together with simple sensors to
implement SLAM using a grid based technique. Unlike traditional techniques, the grid base
technique is less sensitive to the precision of the sensors and will be able to better determine
the position of the robot in places with fewer landmarks. Overall, the final product of the
project is able to explore an indoor environment and produce a fairly accurate map
|
|---|