Simultaneous localisation and mapping of a mobile robot in a static environment

Abstract

The Simultaneous Localisation And Mapping (SLAM) problem in modern‐day robotics and AI is the ability of an autonomous robot to map out its environment whilst simultaneously keeping track of its global position and its movement as it traverses through the space. The robot starts from an unknown location and takes decisions on where it needs to go while adapting according to the environment to accurately localise itself and the objects around it. There are many solutions to solve this problem, such as Graph Simultaneous Localisation And Mapping (Graph‐SLAM), Extended Kalman Filter Simultaneous Localisation And Mapping (EKF‐SLAM), Markov‐SLAM, Iterative Closest Point Simultaneous Localisation And Mapping (ICP‐SLAM) and Fast Simultaneous Localisation And Mapping (Fast‐SLAM) amongst others. In this project, we explore SLAM in the context of a static environment by choosing four SLAM algorithms, analysing how they work theoretically, as well as compare their benefits and drawbacks. The algorithms chosen are Graph‐SLAM, EKF‐SLAM, Fast‐SLAM, and ICP‐SLAM. This was done by looking at previous researchers and their work regarding this problem. Next, we perform algorithm simulations using the Python programming language and a Raspberry Pi 3 B+ for processing to introduce a practical component and compare each algorithm in terms of speed, efficiency, and accuracy. We also compare these results to previous literature, where we observe that all but the Fast‐SLAM algorithm are consistent with other researchers’ findings, with the latter being slower than the other three. Lastly, we take inspiration from the four algorithms tested and develop a practical application of our own SLAM algorithm for a mobile robot. We use the Elegoo Smart Robot Car Kit V4.0, an Arduino Uno as an Input/Output Subsystem and the Raspberry Pi 3 B+ for the processing of the data received. We also use an ultrasonic sensor to detect obstacles around the robot, which is then used for the mapping portion of the robot’s environment, and an accelerometer/gyroscope component to measure the robot’s bearing and orientation. The robot moves and turns using Differential Steering and all sensor readings are processed by the Arduino Uno before being sent to the Raspberry Pi for data visualisation. The developed solution was tested using a variety of environments with different obstacle configurations and starting points for the robot. After gathering the results which consisted of the time taken, as well as the computer resources expended, we compare our map layouts with those obtained by other researchers to verify their consistency and performance.