This paper aims to develop a multi-rotor-based visual tracker for a specified moving object. Visual object-tracking algorithms for multi-rotors are challenging due to multiple issues such as occlusion, quick camera motion, and out-of-view scenarios. Hence, algorithmic changes are required for dealing with images or video sequences obtained by multi-rotors. Therefore, we propose two approaches: a generic object tracker and a class-specific tracker. Both tracking settings require the object bounding box to be selected in the first frame. As part of the later steps, the object tracker uses the updated template set and the calibrated RGBD sensor data as inputs to track the target object using a Siamese network and a machine-learning model for depth estimation. The class-specific tracker is quite similar to the generic object tracker but has an additional auxiliary object classifier. The experimental study and validation were carried out in a robot simulation environment. The simulation environment was designed to serve multiple case scenarios using Gazebo. According to the experiment results, the class-specific object tracker performed better than the generic object tracker in terms of stability and accuracy. Experiments show that the proposed generic tracker achieves promising results on three challenging datasets. Our tracker runs at approximately 36 fps on GPU.

State-of-the-art approaches for localization and mapping are based on local features in images. Along with these features, modern augmented and mixed-reality devices enable building a mesh of the surrounding space. Using this mesh map, we can solve the problem of cross-device localization. This approach is independent of the type of feature descriptors and SLAM used onboard the AR/MR device. The mesh could be reduced to the point cloud that only takes vertices. We analyzed and compared different point cloud registration methods applicable to the problem. In addition, we proposed a new pipeline Feature Inliers Graph Registration Approach (FIGRA) for the co-localization of AR/MR devices using point clouds. The comparative analysis of Go-ICP, Bayesian-ICP, FGR, Teaser++, and FIGRA shows that feature-based methods are more robust and faster than ICP-based methods. Through an in-depth comparison of the feature-based methods with the usual fast point feature histogram and the new weighted height image descriptor, we found that FIGRA has a better performance due to its effective graph-theoretic base. The proposed pipeline allows one to match point clouds in complex real scenarios with low overlap and sparse point density.