DARE-SLAM: Degeneracy-Aware and Resilient Loop Closing in Perceptually-Degraded Environments

Abstract

Enabling fully autonomous robots capable of navigating and exploring largescale, unknown and complex environments has been at the core of robotics research for several decades. A key requirement in autonomous exploration is building accurate and consistent maps of the unknown environment that can be used for reliable navigation. Loop closure detection, the ability to assert that a robot has returned to a previously visited location, is crucial for consistent mapping as it reduces the drift caused by error accumulation in the estimated robot trajectory. Moreover, in multi-robot systems, loop closures enable merging local maps obtained by a team of robots into a consistent global map of the environment. In this paper, we present a degeneracy-aware and drift-resilient loop closing method to improve place recognition and resolve 3D location ambiguities for simultaneous localization and mapping (SLAM) in GPS-denied, large-scale and perceptually-degraded environments. More specifically, we focus on SLAM in subterranean environments (e.g., lava tubes, caves, and mines) that represent examples of complex and ambiguous environments where current methods have inadequate performance. The first contribution of this paper is a degeneracyaware lidar-based SLAM front-end to determine the observability and level of geometric degeneracy in an unknown environment. Using this crucial capability, ambiguous and unobservable areas in an unknown environment are determined and excluded from the search for loop closures to avoid distortions of the entire map as the result of spurious or inaccurate loop closures. The second contribution of this paper is a drift-resilient loop closing pipeline that exploits the salient 2D and 3D features extracted from lidar point cloud data to enable a

NeBula: Quest for Robotic Autonomy in Challenging Environments; TEAM CoSTAR at the DARPA Subterranean Challenge

“…The multi-modal architecture of LAMP's loop closure module (Figure 12) enables a robust and reliable system through the use of different sensing modalities. These loop closure sensing modalities include using lidar data (Ebadi et al, 2020), visual data (Rosinol et al, 2020) and semantic data (Ebadi et al, 2021). Figure 12: LAMP Architecture.…”

“…(b) Environmental Landmark Factors: Existing features in the environment such as signs, salient objects and the shape of junctions (e.g. (Ebadi et al, 2021)) can be used as landmarks. For example, we use observations of specific objects, such as backpacks and fire extinguishers (called artifacts in SubT), with sets of range-bearing observations (dashed black lines in Figure 13) from the artifact relative-localization module (Section 7).…”

Towards Collaborative Simultaneous Localization and Mapping: a Survey of the Current Research Landscape

“…To that end, compact and robust global point cloud descriptors (Uy and Lee, 2018) can be relied upon to compare point clouds for place recognition. Other approaches extract features from the point cloud that can serve for place recognition while providing initial guesses for later geometric alignments (Ebadi et al, 2021), or even directly compute loop closure measurements (Dubé et al, 2017a). While the classical Iterative Closest Point method (Besl and McKay, 1992) is still commonly used in single robot SLAM to compute relative pose measurements between two matching point clouds, it is not well suited for multi-robot operation due to its reliance on a good initial guess, which is usually not available between the robots' local maps.…”

“…While the classical Iterative Closest Point method (Besl and McKay, 1992) is still commonly used in single robot SLAM to compute relative pose measurements between two matching point clouds, it is not well suited for multi-robot operation due to its reliance on a good initial guess, which is usually not available between the robots' local maps. Therefore, a common solution is to use submaps matching for both stereo cameras (Schuster et al, 2015;Schulz et al, 2019) and lidars (Dubé et al, 2017b;Dubois et al, 2020b;Ebadi et al, 2021). During this process, multiple laser scans or 3D point clouds are clustered into submaps which can in turn be registered more efficiently.…”

“…Moreover, many ambitious applications remain for multi-robot systems, such as the exploration of other planets (Vitug, 2021;Ebadi et al, 2021). To reach those moonshot goals, ongoing trends in the research community aim to push the boundaries of multi-robot systems towards increasingly larger teams, or swarms of robots (Beni, 2004;Kegeleirs et al, 2021), which potentially allow parallel operations that are more efficient and versatile.…”