Olusanya Y. Agunbiade,∗ Mziwoxolo Mayedwa,∗ and Awosejo Oluwaseun Johnson∗∗
Independent vehicle, dynamic condition, lighting condition, kidnapping event and shadow
The issue of simultaneous localization and mapping (SLAM) has been thoroughly investigated in robotics. Its influence on independent robot navigation attracted scholars. Over decades, a variety of approaches have been suggested to handle the SLAM problem with commendable success, however there are a variety of factors that can reduce the efficiency of the SLAM techniques. Environmental elements, such as lighting conditions, shadow, dynamic (non-static) conditions, kidnapping event, computational complexity, and shadows are some of these concerns (factors). These challenges (factors) produce inconsistencies, which might result in execution that yields undesirable results. In an attempt to overcome these challenges, a unique SLAM approach identified as DIK-SLAM has been presented. The dynamic illumination and kidnapping (DIK) SLAM technique is an improved version of the Monte- Carlo (MCL) SLAM algorithm that incorporates filtering techniques and various adjustments to increase the reliability and considering computational cost. The normalised differences index (NDI) is the filtering approach used by the DIK-SLAM to eliminate shadow. To minimise the effect of light intensity, filters like specular-to-diffuse and dark channel models were also applied to the DIK-SLAM. Given that the computational cost is a consideration, these filtering techniques are running concurrently. In addressing the kidnapping problem and the dynamic (non-static) environment, respectively, the revised MCL algorithm founded on grid map and initial localisation approach was presented to create the DIK-SLAM. In this article, the SLAM algorithms were evaluated using a publicly released dataset (TUM-RGBD). The MATLAB simulation software was used to conduct the test, and results were evaluated quantitatively. Thus, comparing the DIK-SLAM, the traditional MCL algorithm, and other SLAM approaches accessible in the literature, experimental results showed that the DIK-SLAM performed better because for most of the trajectory evaluation it attained lower error. The DIK-SLAM technique presented in this paper has the ability to support independent movement, route planning, and exploration while minimising the robot failure rate, injuries and accidents to humans.
Important Links:
Go Back
