Deep learning with attention model applied to occupation grid analysis
Published : 13 December 2016
This topic falls in the context of the development of autonomous vehicles, drones, and robotics.
The environment of the vehicle is described in an occupation grid, each cell of the grid containing the probability of occupation by an object. This grid is updated over time with sensors data.
Higher-level algorithms, like path planning or collision avoidance, think in terms of objects described by their path, speed, and nature. It is thus mandatory to get these objects from individual grid cells, with clustering, classification, and tracking.
Most of the previous publications on this topic comes from the context of vision processing, many of them using deep learning. They show a big computational complexity, and do not benefit from occupation grids specific characteristics (lack of textures, a priori knowledge of areas of interest…). In this PhD, we want to explore new techniques, tailored to occupation grids, and more compatible with embedded and low cost implementation.
The purpose of this thesis is, starting from a fusion-based occupation grid, to get the contained objects, including their position, speed vector, and nature, by using attention-based artificial neuron networks.