Generative Simultaneous Localization and Mapping (G-SLAM)

TL;DR

G-SLAM introduces a probabilistic method for robot mapping and localization using scattered points and particle filters, demonstrating adaptive point repositioning and convergence around obstacles in real-world scenarios.

Contribution

The paper presents a novel G-SLAM method that combines probabilistic mapping with adaptive point repositioning for improved robot localization.

Findings

Effective obstacle mapping with scattered points

Adaptive repositioning improves convergence around obstacles

Validated with real vehicle experiments

Abstract

Environment perception is a crucial ability for robot's interaction into an environment. One of the first steps in this direction is the combined problem of simultaneous localization and mapping (SLAM). A new method, called G-SLAM, is proposed, where the map is considered as a set of scattered points in the continuous space followed by a probability that states the existence of an obstacle in the subsequent point in space. A probabilistic approach with particle filters for the robot's pose estimation and an adaptive recursive algorithm for the map's probability distribution estimation is presented. Key feature of the G-SLAM method is the adaptive repositioning of the scattered points and their convergence around obstacles. In this paper the goal is to estimate the best robot trajectory along with the probability distribution of the obstacles in space. For experimental purposes a four wheel rear drive car kinematic model is used and results derived from real case scenarios are discussed.