It is often necessary for drones to complete delivery, photography, and rescue in the shortest time to increase efficiency. Many autonomous drone races provide platforms to pursue algorithms to finish races as quickly as possible for the above purpose. Unfortunately, existing methods often fail to keep training and racing time short in drone racing competitions. This motivates us to develop a high-efficient learning method by imitating the training experience of top racing drivers. Unlike traditional iterative learning control methods for accurate tracking, the proposed approach iteratively learns a trajectory online to finish the race as quickly as possible. Simulations and experiments using different models show that the proposed approach is model-free and is able to achieve the optimal result with low computation requirements. Furthermore, this approach surpasses some state-of-the-art methods in racing time on a benchmark drone racing platform. An experiment on a real quadcopter is also performed to demonstrate its effectiveness.

In mobile Augmented Reality (AR) applications, relocalization plays a very important role in supporting persistence and multiple user interaction. Traditional methods usually utilize single-frame and integrated into the loop closure procedure in simultaneous localization and mapping (SLAM). However, indoor scenes include many challenges: similar patterns such as stairs and kitchen corners, textureless places such as white walls. To overcome these challenges, we propose a novel sequence-based relocalization method for mobile AR devices. Our method utilizes the commodity depth sensors such as ToF cameras. We firstly generate a sequence of depth maps with corresponding poses tracked via Visual Inertial Odometry (VIO). Based on keyframe-based visual relocalization, then choose a good subset of posed depth maps to verify and refine the pose. Results show that our method improves relocalization accuracy compared with simple sequence-based relocalization.