Implementations of efficient neural rendering pipeline to generate view-consistency images and pseudo-labels for replay-based continual learning for semantic segmentation
The focus will be to assess people navigation behavior around the robot by extracting trajectories and motions. I am working on a novel detecting, tracking, and motion profile extraction pipeline on lidar and camera data.
Implementation of a working, simple, monocular visual odometry (VO) pipeline in Matlab
Proposed a simplified pipeline of last-centimeter drone delivery towards window/balcony with vision-based fiducial marker detection and collision prevention under rigorous test
We programed based on [Crazyflie 2.1](https://www.bitcraze.io/products/crazyflie-2-1/) to find and precisely land on a platform with height of 10 cm by utilizing z reading from [flow deck](https://www.bitcraze.io/products/flow-deck-v2/). Additionally, We also utilized sensor readings from [multi-ranger deck](https://www.bitcraze.io/products/multi-ranger-deck/) to avoid the obstacles presented in the environment.
The rapid development of autonomous driving and mobile mapping calls for off-the-shelf LiDAR SLAM solutions that are adaptive to LiDARs of different specifications on various complex scenarios. To this end, we propose MULLS, an efficient, low-drift, …
Object tracking has been broadly applied in unmanned aerial vehicle (UAV) tasks in recent years. However, existing algorithms still face difficulties such as partial occlusion, clutter background, and other challenging visual factors. Inspired by the …
Improved the existing trackers on overall performance in challenging UAV scenarios with high operational efficiency
In recent years, the correlation filter (CF)-based method has significantly advanced in the tracking for unmanned aerial vehicles (UAVs). As the core component of most trackers, CF is a discriminative classifier to distinguish the object from the …