Robotic talks in the Guangzhou Institute of Advanced Tecnology

Javier Finat will lecture two sessions this week about Robotics at the Guangzhou Institute of Advanced Technology (GIAT) located at Haibin Rd., 1121, Nansha, Guangzhou.

In the first talk, he is going to introduce the main elements involved in the localization, navigation and trajectory planning of mobile robots. Information coming from different types of sources (sensors, cameras and LiDAR) must be fused to predict more accurate poses and trajectories of the robots. Geometric methods and statistical approaches for sampling can be combined in the classical pipeline: sample, filter, detect, describe, predict, modify and interact. However, it is mandatory to handle noise in real-time data to deal with uncertainty. The goal of this talk is to discuss analytical and statistical approaches to these problem with the researches of the GIAT.

In the second talk, Simultaneous Localization and Mapping (SLAM) is going to be addressed. SLAM is field of robotics to determine the trajectory of a camera at the same time that the map of the scene is being reconstructed. The field combines techniques from other areas like Structure from Motion (SFM) and Multiple View Geometry (MVG) to estimate the trajectory of the robot. Depending on the number and type of sensors the problem is more or less difficult to tackle. Indirect approaches relies on featured key points extracted from pixels that makes the system robust to noise but requires more time per frame. Direct approaches minimize the photometric error between consecutive frames to determine relative rigid transformations between frames. This talk will discuss these topics from the beginnings at the nineties until latest approaches based on Deep Learning.

Both talks will take place next weeks as the kick off of a collaboration between MoBiVAP and the GIAT.