New Monte Carlo Localization Using Deep Initialization: A Three-Dimensional LiDAR and a Camera Fusion Approach

Abstract

Fast and accurate global localization of autonomous ground vehicles is often required in indoor environments and GPS-shaded areas. Typically, with regard to global localization problem, the entire environment should be observed for a long time to converge. To overcome this limitation, a new initialization method called deep initialization is proposed and it is applied to Monte Carlo localization (MCL). The proposed method is based on the combination of a three-dimensional (3D) light detection and ranging (LiDAR) and a camera. Using a camera, pose regression based on a deep convolutional neural network (CNN) is conducted to initialize particles of MCL. Particles are sampled from the tangent space to a manifold structure of the group of rigid motion. Using a 3D LiDAR as a sensor, a particle filter is applied to estimate the sensor pose. Furthermore, we propose a re-localization method for performing initialization whenever a localization failure or the situation of robot kidnapping is detected. Either the localization failure or the kidnapping is detected by combining the outputs from a camera and 3D LiDAR. Finally, the proposed method is applied to a mobile robot platform to prove the method's effectiveness in terms of both the localization accuracy and time consumed for estimating the pose correctly.

Vision-Based UAV-UGV Collaboration for Autonomous Construction Site Preparation

“…For example, using an onboard laser range finder and matching the data requires slow base motion, thus increasing the time to finish a task [35]. In other cases, such as using symmetric maps without relative measurements, the localization may not converge at all [38]. Therefore, research can benefit from the use of an additional agent providing accurate online localization by monitoring the UGV.…”

A Geodetic Normal Distribution Map for Long-Term LiDAR Localization on Earth

“…Given the GND map m GN D and measured points z t = {z 1 t , z 2 t , • • • , z j t } at time t in the vehicle coordinate, the map-matching problem is formulated as estimating the probability density of the vehicle pose p(x t |z t , m GN D ), where the point z j t is composed of threedimensional position information (x, y, z) and the vehicle pose x t is composed of six elements: longitude, latitude, height, roll, pitch, and yaw. The map-matching problem for localization can be categorized into two types: Monte-Carlo [48], [49] and optimization approaches [50], [51]. The two types of map-matching approaches are described in the next sections.…”

Highlighted Map for Mobile Robot Localization and Its Generation Based on Reinforcement Learning

“…(lines 3-7), the particles are randomly sampled based on the ratio of the likelihood w (i) t (lines [8][9][10][11]. This sampling process is called resampling.…”

“…The method in [7] combined MCL with a laser scan matching algorithm to inherit the stability of MCL and the high accuracy of the scan matching. A particle initialization method by a convolutional neural network was proposed in [8]. The major target of MCL is a wheeled robot, but it has also been applied to an underwater robot [9], an unmanned aerial vehicle [10], and the field of wireless sensor networks [11].…”