Collaborative localization of aerial and ground robots through elevation maps

Abstract

Collaboration between aerial and ground robots can benefit from exploiting the complementary capabilities of each system, thereby improving situational awareness and environment interaction. For this purpose, we present a localization method that allows the ground robot to determine and track its position within a map acquired by a flying robot. To maintain invariance with respect to differing sensor choices and viewpoints, the method utilizes elevation maps built independently by each robot's onboard sensors. The elevation maps are then used for global localization: specifically, we find the relative position and orientation of the ground robot using the aerial map as a reference. Our work compares four different similarity measures for computing the congruence of elevation maps (akin to dense, image-based template matching) and evaluates their merit. Furthermore, a particle filter is implemented for each similarity measure to track multiple location hypotheses and to use the robot motion to converge to a unique solution. This allows the ground robot to make use of the extended coverage of the map from the flying robot. The presented method is demonstrated through the collaboration of a quadrotor equipped with a downward-facing monocular camera and a walking robot equipped with a rotating laser range scanner. Abstract-Collaboration between aerial and ground robots can benefit from exploiting the complementary capabilities of each system, thereby improving situational awareness and environment interaction. For this purpose, we present a localization method that allows the ground robot to determine and track its position within a map acquired by a flying robot. To maintain invariance with respect to differing sensor choices and viewpoints, the method utilizes elevation maps built independently by each robot's onboard sensors. The elevation maps are then used for global localization: specifically, we find the relative position and orientation of the ground robot using the aerial map as a reference. Our work compares four different similarity measures for computing the congruence of elevation maps (akin to dense, image-based template matching) and evaluates their merit. Furthermore, a particle filter is implemented for each similarity measure to track multiple location hypotheses and to use the robot motion to converge to a unique solution. This allows the ground robot to make use of the extended coverage of the map from the flying robot. The presented method is demonstrated through the collaboration of a quadrotor equipped with a downward-facing monocular camera and a walking robot equipped with a rotating laser range scanner.

Learning Ground Traversability From Simulations

“…In order to validate whether the classifier is effective in a real scenario, we built the Bars environment (Figure 1) which features two horizontal wooden bars with a rectangular cross section as obstacles. One bar is 6 cm deep whereas [30], Sullens [30], Gravelpit [30] (one of three areas extracted from the same scenario), ETH-ASL [31] and Slope (acquired area corresponds to the blue outline). the other is 8 cm deep.…”

Overview obstacle maps for obstacle‐aware navigation of autonomous drones

“…REMODE has since been utilized on data acquired with drones to generate dense depth maps in real time for various research projects: creating medium-sized maps in an offboard ground-station by streaming the acquired data (Faessler et al, 2016), the creation of dense maps onboard (Forster, Faessler, Fontana, Werlberger, & Scaramuzza, 2015) by restricting their size to a relatively small 2.5D fixed-size grid-map around the robot (Fankhauser, Bloesch, Gehring, Hutter, & Siegwart, 2014), and the feasibility of sharing the 2.5D map acquired by the drone to guide a ground robot. Regarding the latter and still using REMODE, a mobile robot plans and executes trajectories in rough terrain in a small area mapped by a drone (Delmerico, Mueggler, Nitsch, & Scaramuzza, 2017), by training a terrain classifier on-the-fly (Delmerico, Giusti, Mueggler, Gambardella, & Scaramuzza, 2016), and a legged robot and the drone achieve localization on the same global coordinate frame in Käslin et al (2016). In Lynen et al (2015), an efficient indexed nearest-neighbor search to achieve image-based relocalization on a prebuilt map is proposed, where the map is obtained using standard SfM techniques with its scale recovered using IMU data and the recursive nonlinear filtering approach OKVIS (Agarwal et al, 2012;Leutenegger et al, 2013Leutenegger et al, , 2015, and a VIO method for local pose tracking inspired on the multi-state constraint kalman filter (MSCKF) (Mourikis et al, 2009) is used.…”

Supervised morphogenesis: Exploiting morphological flexibility of self-assembling multirobot systems through cooperation with aerial robots

“…Nevertheless, they have been successfully applied to solve various tasks such as to cooperatively map obstacles in large areas [32], to augment the view of a moving groundbased robot with aerial images [33,34], and to cooperatively track a moving target [35]. In more recent work, Käslin et al [36] presented a localization method based on elevation maps for ground robots. The method is independent of sensors and allows a ground robot to find its relative position and orientation within the reference map provided by an aerial robot without relying on GPS.…”