Abstract
In this work, we propose the use of Ground Penetrating Radar (GPR) for roverlocalization on Mars. Precise pose estimation is an important task for mobilerobots exploring planetary surfaces, as they operate in GPS-deniedenvironments. Although visual odometry provides accurate localization, it iscomputationally expensive and can fail in dim or high-contrast lighting. Wheelencoders can also provide odometry estimation, but are prone to slipping on thesandy terrain encountered on Mars. Although traditionally a scientificsurveying sensor, GPR has been used on Earth for terrain classification andlocalization through subsurface feature matching. The Perseverance rover andthe upcoming ExoMars rover have GPR sensors already equipped to aid in thesearch of water and mineral resources. We propose to leverage GPR to aid inMars rover localization. Specifically, we develop a novel GPR-based deeplearning model that predicts 1D relative pose translation. We fuse our GPR poseprediction method with inertial and wheel encoder data in a filtering frameworkto output rover localization. We perform experiments in a Mars analogenvironment and demonstrate that our GPR-based displacement predictions bothoutperform wheel encoders and improve multi-modal filtering estimates inhigh-slip environments. Lastly, we present the first dataset aimed at GPR-basedlocalization in Mars analog environments, which will be made publicly availableat https://umfieldrobotics.github.io/marslgpr.