Localization Limitations of ARCore, ARKit, and Hololens in Dynamic Large-scale Industry Environments

Feigl, Tobias; Porada, Andreas; Steiner, Steve J.; Löffler, Christoffer; Mutschler, Christopher; Philippsen, Michæl

doi:10.5220/0008989903070318

Cited by 39 publications

(30 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Environmental understanding: This process involves grouping the characteristic points, which are identified by computer vision algorithms, and belong to common horizontal or vertical surfaces (e.g., tables or walls). The recognized surfaces from this process, are then defined as planes with distinct boundaries on which virtual 3D models can be placed, making them look like part of the real world (Feigl, Tobias et al 2020).…”

Section: __________________________ * Corresponding Authormentioning

confidence: 99%

Aligning Positioning and Navigation of Smart Mobile Devices to Ar/Vr Environments

Kaimaris

Roustanis

Κάρολος

et al. 2021

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Augmented Reality/Virtual Reality (AR/VR) technology offers novel and excited experiences to visitors of cultural sites. Extended recent technological advances in smart communication devices made the connection of the real to the virtual worlds more affordable and effective. In spite of the rapid development and availability of the AR/VR applications to cultural heritage, there still exist gaps and challenges in accurate positioning and navigation of visitors’ smart devices in both out-doors archaeological spaces as well as in in-doors museum spaces. Recent technology smart devices embody GNSS positioning sensors, which in best cases, provide 1-meter positioning accuracies in open spaces, while are unable to work in-doors. Such accuracies and restrictions lead to gaps in fetching visitor’s position in AR environments and mis-positions with regard to the rest AR/VR objects as visualized in smart screens, while the problem gets worse when dealing with real-time videoing, when elapse times and refresh rates lead to additional mispositionings. In this paper, effort is made to present and evaluate the most recent and widely used tools for positioning and navigation of smart devices in in- and out-doors CH sites.

show abstract

Section: __________________________ * Corresponding Authormentioning

confidence: 99%

Aligning Positioning and Navigation of Smart Mobile Devices to Ar/Vr Environments

Kaimaris

Roustanis

Κάρολος

et al. 2021

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…the presence of moving others). Feigl et al [13] assessed the localisation accuracy of ARCore, ARKit and Hololens devices, finding that "out of the box, these AR systems are far from useful even for normal motion behaviour", with an average error of approximately 17 m per 120 m when assessed in a largescale industry environment 60 m+ traversals). And Duque [10] assessed the Oculus Rift CV1 outside-in tracking, finding distance error of approximately 1.7 cm.…”

Section: Assessing the Accuracy Of Slam Trackingmentioning

confidence: 99%

“…Consequently, determining whether a given SLAM-tracked device will work for a particular environment/use case is difficult to answer without some form of assessment insitu. Problematically, common to both these papers was the use of external optical tracking as a ground truth for benchmarking (e.g., ARTTRACK cameras in [13], outside-in tracking cameras in [10]). This significantly increases the cost and complexity of insitu assessments in different spaces, with optical tracking solutions capable of tracking devices over large spaces costing significant sums and requiring the installation of hardware infrastructure.…”

Section: Assessing the Accuracy Of Slam Trackingmentioning

confidence: 99%

“…This latter point is, however, highly prescient -the accuracy of our alignment techniques is dependent both on the application of each technique, and also the underlying robustness of the SLAM tracking implementation. We could trivially perform a two-point alignment with 0.01m worst case theoretical error at the edges of the play area, however the error in reality is also contingent on the accuracy of the SLAM tracking solution in a given environmentand this can vary markedly depending on said environment and the sensing available on the device [13]. Consequently, for large scale co-location in particular, we need the ability to assess the accuracy of any arbitrary combination of MR device and environment, so that we can determine a) whether safe operation is possible, and b) what safety margins need to be incorporated into the experience (e.g., to prevent occluded VR users from accidentally colliding).…”

Section: Assessing Slam Robustnessmentioning

confidence: 99%

See 1 more Smart Citation

A Quest for Co-Located Mixed Reality: Aligning and Assessing SLAM Tracking for Same-Space Multi-User Experiences

McGill

Gugenheimer

Freeman

2020

26th ACM Symposium on Virtual Reality Software and Technology

View full text Add to dashboard Cite

Current solutions for creating co-located Mixed Reality (MR) experiences typically rely on platform-specific synchronisation of spatial anchors or Simultaneous Localisation and Mapping (SLAM) data across clients, often coupled to cloud services. This introduces significant costs (in development and deployment), constraints (with interoperability across platforms often limited), and privacy concerns. For practitioners, support is needed for creating platformagnostic co-located MR experiences. This paper explores the utility of aligned SLAM solutions by 1) surveying approaches toward aligning disparate device coordinate spaces, formalizing their theoretical accuracy and limitations; 2) providing skeleton implementations for audience-based, small-scale and large-scale co-location using said alignment approaches; and 3) detailing how we can assess the accuracy and safety of 6DoF/SLAM tracking solutions for any arbitrary device and dynamic environment without the need for an expensive ground truth optical tracking, by using trilateration and a $30 laser distance meter. Through this, we hope to further democratise the creation of cross-platform co-located MR experiences. CCS CONCEPTS • Human-centered computing → Virtual reality; • Computing methodologies → Mixed / augmented reality;

show abstract

“…Yet, optical-flow based relative navigation systems are highly sensitive for major drifts. See [ 30 ] for a detailed accuracy evaluation of

and

platforms. The main causes of such drifts are: (a) Compass drift due to metal interference or local magnets (see Figure 6 for an example of a major error due to compass drift).…”

Section: Improved Algorithmmentioning

confidence: 99%

STEPS: An Indoor Navigation Framework for Mobile Devices

Landau

Ben‐Moshe

2020

Sensors

View full text Add to dashboard Cite

This paper presents a vision-based navigation system designed for indoor localization. The suggested framework works as a standalone 3 D positioning system by fusing a sophisticated optical-flow pedometry with map constrains using an advanced particle filter. The presented method requires no personal calibration and works on standard smartphones with relatively low energy consumption. Field experiments on Android smartphones show that the expected 3 D error is about 1–2 m in most real-life scenarios.

show abstract

Localization Limitations of ARCore, ARKit, and Hololens in Dynamic Large-scale Industry Environments

Cited by 39 publications

References 23 publications

Aligning Positioning and Navigation of Smart Mobile Devices to Ar/Vr Environments

Aligning Positioning and Navigation of Smart Mobile Devices to Ar/Vr Environments

A Quest for Co-Located Mixed Reality: Aligning and Assessing SLAM Tracking for Same-Space Multi-User Experiences

STEPS: An Indoor Navigation Framework for Mobile Devices

Contact Info

Product

Resources

About