Detection and Tracking of General Movable Objects in Large Three-Dimensional Maps

Abstract

IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.Abstract-This paper studies the problem of detection and tracking of general objects with long-term dynamics, observed by a mobile robot moving in a large environment. A key problem is that due to the environment scale, it can only observe a subset of the objects at any given time. Since some time passes between observations of objects in different places, the objects might be moved when the robot is not there. We propose a model for this movement in which the objects typically only move locally, but with some small probability they jump longer distances, through what we call global motion. For filtering, we decompose the posterior over local and global movements into two linked processes. The posterior over the global movements and measurement associations is sampled, while we track the local movement analytically using Kalman filters. This novel filter is evaluated on point cloud data gathered autonomously by a mobile robot over an extended period of time. We show that tracking jumping objects is feasible, and that the proposed probabilistic treatment outperforms previous methods when applied to real world data. The key to efficient probabilistic tracking in this scenario is focused sampling of the object posteriors.

Uncertainty-based localization in a topological robot navigation system

“…Binary object classification identifies objects as movable or static. Most of the approaches that use binary classification are meant to map the static objects and discard the movable ones [112,167,26]. In order to replicate this behavior, only the objects that have been labeled as active or unknown for all of the mapping sessions are included in the resulting map since they are supposed to be static.…”

“…An object-based pose graph is developed in [167], where the most up to date representation of the environment is maintained by merging new objects and removing old ones. In [26], Bore et al propose tracking of objects in 3D maps while mantaining an updated representation of the environment over time. When an object has disappeared from its mapped position, the system looks for it until its new location is found.…”

Object-Based Pose Graph for Dynamic Indoor Environments

“…An object-based pose graph is developed in [22], where the most up to date representation of the environment is maintained by merging new objects and removing old ones. In [23] tracking of objects is performed in 3D maps. When an object has disappeared from its mapped position, the system looks for it until its new location is found.…”

“…Binary object classification identifies objects as movable or static. Most of the approaches that use binary classification are meant to map the static objects and discard the movable ones [21], [22], [23]. In order to replicate this behavior, only the objects that have been labeled as active or unknown for all of the mapping sessions are included in the resulting map as they are supposed to be static.…”

Rescan: Inductive Instance Segmentation for Indoor RGBD Scans

“…Other work has aimed at life-long scene understanding using data captured with actively controlled sensors [15,29,39,49]. For example, several algorithms proposed in the STRANDS project [23] process the scenes observed from a repeated set of views [2,6,41]. Others focus on controlling camera trajectories to acquire the best views for object modeling [13,15] and/or change detection [1].…”