Exploiting Moving Objects: Multi-Robot Simultaneous Localization and Tracking

Abstract

Cooperative localization has been proved to effectively outperform single-robot localization. While most of the state-of-the-art multi-robot localization systems either treat moving objects as outliers or accomplish moving object tracking separately from localization, we argue that augmenting moving objects into the localization estimation can further enhance localization performance and is indeed the key to solve several localization challenges such as insufficient map features, no map features, and symmetric maps. In this paper, a multi-robot simultaneous localization and tracking (MR-SLAT) algorithm based on the extended Kalman filter is proposed, and multiple hypothesis tracking (MHT) is integrated into MR-SLAT for dealing with challenging data association issues. The proposed approach is verified in two scenarios: the NAO humanoid robots equipped with cameras and WiFi are used in the RoboCup scenario and the robotic vehicles with laser scanners and dedicated short-range communications (DSRC) are used in the traffic scenario. The experiments with ground truth show that MR-SLAT, by exploiting moving objects, is superior to single-robot localization and cooperative localization in challenging scenarios. Ample experimental and simulation results demonstrate the effectiveness of exploiting moving objects and the generality and feasibility of the proposed MR-SLAT algorithm.Note to Practitioners: Abstract-To apply robot systems in realworld dynamic scenes, the capability of dealing with moving objects is necessary and critical. The key insight of this work is that by exploiting moving objects, many practical localization cases that are challenging for approaches which filter out or separately deal with moving objects can be solved. The main observation is that in the multi-robot scenario, the commonly observed moving objects actually contain valuable information that helps localization. The proposed MR-SLAT algorithm exploits the mutual benefits between localization and tracking by maintaining the correlations between teammate robots and moving objects through the augmented state and thus further improves localization accuracy and reliability. In addition, in the proposed system, data association, which is one of the most critical parts for robot systems to work successfully in practical challenging scenes, is tackled by integrating MHT, and the feasibility is experimentally demonstrated and analyzed. The applicability of the proposed system and the concept of exploiting moving objects are verified through ample experiments in two different scenarios with different platforms and sensors. The manuscript, along with the proposed MR-SLAT approach, accordingly makes contributions to practical multi-agent applications such as intelligent transportation systems.Index Terms-Cooperative systems, multi-sensor systems, robot sensing systems.

SLAM in Dynamic Environments: A Deep Learning Approach for Moving Object Tracking Using ML-RANSAC Algorithm

“…Additionally, the assumption of tracking only one target in [9,10,32,33], makes those impossible to be implemented in real dynamic environments. The algorithms of [8,10,11,12,32,36] are also locally applicable and require high computational effort and memory in large environments. The classical object detection methods are applied for object detection in these algorithms, which are not suitable for real environments and are prone to errors in unstructured dynamic environments.…”

“…Azim and Aycard [35] investigated detection, classification, and tracking of moving objects using a 3D range laser in dynamic outdoor environment, where GNN technique was employed for data association.However, the applied data association technique is not proper for dynamic environments and will fail, especially in proximity and occlusion situations. An EKF-based algorithm for multi-robot simultaneous localization and tracking, using MHT, was proposed by [11] to solve data association problem. The advantage of multi sensor fusion and classification at detection level was studied in [36] by using LIDAR, radar, and camera as the main sensors for the moving object detection and tracking.…”

“…Although such methodologies are effective in static environments, they fail in realistic scenarios whereby moving objects are present. In recent years, however, researchers have attempted to design SLAM algorithms for unstructured, uncertain, and dynamic environments [8,9,10,11]. In order for an autonomous robot to navigate through an unknown environment; two problems need to be resolved: (i) SLAM problem, which builds and updates the environment map while localizing the robot with respect to that map and (ii) Detection And Tracking of Moving Objects (DATMO) nearby the robot and estimating their future behavior.…”

Task and Motion Coordination for Heterogeneous Multiagent Systems With Loosely Coupled Local Tasks

“…Agent a i can exchange messages directly with any agent a j ∈ N ai t . Inspired by the idea of backbone network [5], we define the following twolayer communication network: Definition 1. The communication network C(t) = (N , E(t)) at time t > 0, where N is the set of nodes and E(t) ⊆ N × N is the set of edges, which has two layers: E(t) = E 1 (t)∪E 2 (t):…”

“…We show that the proposed scheme guarantees the satisfaction of all local tasks and potential agent failures can be recovered. Some potential applications can be found in [5].…”