Many planetary robotics applications require (semi-)autonomous rover operation for which localization is essential. During the ESA-ESRIC Space Resources Challenge in 2021 and more recently AMADEE-24 in Armenia in March 2024, we put a methodology where a mobile system first distributes a set of Ultra-Wideband (UWB) transceivers (anchors) to create a location-aware wireless sensor network (WSN) to the test.

This allows the localization of the robot based on trilateration. More specifically, an STM32 board with Decawave’s DW1000 UWB transceiver IC is distributed with a small 3D printed CO2 powered rocket, see Figure 1 (left). The STM32 microcontroller runs Rodos (Realtime Onboard Dependable Operating System), a real-time operating system for embedded systems. A Rodos-ROS bridge makes the data available to the robot and eases the operation. At the time of launch (Figure 1 (middle)), a spring-loaded striker pierces the cartage inside a sealed launch tube. As a result, the armature is ejected from the launcher tube. A mobile robot with two launcher tubes with attached striker mechanisms, as shown in Figure 1 (right), was tested in Armenia.

The UWB sensors perform asymmetric double- sided two-way ranging (ADS-TWR), from a remaining node on the robot (tag) – a ranging method that uses the delays naturally occurring in signal transmission to determine the distance between two nodes, thus eliminating the need for clock synchronization between the UWB sensors. As the position of the anchors after the launch of the rocket is unclear, common trilateration cannot be used. Therefore, adapted range-only SLAM methods are integrated into the planetary rover software, to preform the position determination. After aligning the coordinate frames of the UWB localization system with the visual odometry of the mobile robot, which is done with an Intel T265 stereo camera, the robot has a globally stable coordinate system for localization. More recently we have also tested a system with three fixed tags on the robot, which enables the use of common trilateration methods to not only determine the distance of the robot to the anchors but also their respective bearing. This data is then used to preform a more robust SLAM algorithm using an Extended Kalman Filter (EKF) to determine the robots pose.

Source: Martin Hesse¹, “UWB Sensor Based Localization of a Planetary Rover”, AMADEE-24 Science Workshop Booklet 2024

¹ Julius-Maximilians-University Würzburg