Map++: Towards User-Participatory Visual SLAM Systems with Efficient Map Expansion and Sharing

TL;DR

This paper develops Map++, an efficient system that functions as a plug-and-play extension, supporting participatory map-building based on existing SLAM algorithms, and addresses a plethora of scalability issues in this participatory map-building system by proposing a set of lightweight, application-layer protocols.

Abstract

Constructing precise 3D maps is crucial for the development of future map-based systems such as self-driving and navigation. However, generating these maps in complex environments, such as multi-level parking garages or shopping malls, remains a formidable challenge. In this paper, we introduce a participatory sensing approach that delegates map-building tasks to map users, thereby enabling cost-effective and continuous data collection. The proposed method harnesses the collective efforts of users, facilitating the expansion and ongoing update of the maps as the environment evolves. We realized this approach by developing Map++, an efficient system that functions as a plug-and-play extension, supporting participatory map-building based on existing SLAM algorithms. Map++ addresses a plethora of scalability issues in this participatory map-building system by proposing a set of lightweight, application-layer protocols. We evaluated Map++ in four representative settings: an indoor garage, an outdoor plaza, a public SLAM benchmark, and a simulated environment. The results demonstrate that Map++ can reduce traffic volume by approximately 46% with negligible degradation in mapping accuracy, i.e., less than 0.03m compared to the baseline system. It can support approximately $2 \times$ as many concurrent users as the baseline under the same network bandwidth. Additionally, for users who travel on already-mapped trajectories, they can directly utilize the existing maps for localization and save 47% of the CPU usage.

Figures (15)

• Figure 1: A User-Participatory SLAM system. Users upload data to contribute to the map (shown in the bottom left corner of the garage) on the server.
• Figure 2: Overview of a vanilla shared-map architecture as discussed in covins. Each user uploads raw data (in the form of keyframes) to the server. The server merges the map from different users and conducts global optimization.
• Figure 3: Map++ Overview. The gray part (how to update) is not part of this work, and citations are given for further reference.
• Figure 4: Illustration of the camera pose, its view cone, and the overlap evaluation between two cones. The sampling points marked as FRESH or REDUNDANT are also included.
• Figure 5: This example illustrates that neighbor map points may fall within the query pose's view cone, but should not be deemed as REDUNDANT. Thus, we cannot simply detect the overlap by evaluating how many neighbor points on the map fall within the view cone of the query pose.