Hi! I'm Eduardo Perdices. I completed a PhD in computer vision at the URJC Robotics Group (Spain). In the past, I studied a Bachelor's degree in computer science engineering (5-year degree) and a Master's degree in telematic and information systems at Rey Juan Carlos university.
I have been working as a researcher for 4 years and currently I'm working as Software Engineer at Star Robotics, designing and programming computer vision applications.
Simultaneous Localization and Mapping (SLAM) using computer vision has achieved a huge breakthrough in the last years. Since MonoSLAM and PTAM approaches where released, many other algorithms have been developed based on these works.
Visual SLAM can be obtained used one single camera, stereo cameras or depth cameras. I have developed a monocular visual SLAM algorithm called SD-SLAM designed to run in devices with low computational capabilities, such as drones or mobile phones. Source code can be obtained here
The next video shows its behavior using Tum RGB-D dataset:
I am developing augmented reality applications for Android mobile phones. This field is related to Visual SLAM, since augmented reality highly depends on device's localization. However, we can also employ other available sensors, such as IMUs, increasing localization accuracy.
Artificial Intelligence tries to imitate the human brain to process data and make decisions. Since 2012, when AlexNet was published, Convolutional Neural Networks (CNN) have been employed to process images and recognize objects.
I use CNN for 3D objects recognition and I am looking for ways to employ deep learning for SLAM purposes.
© 2020 Eduardo Perdices. All rights reserved
