Low-cost time-of-flight-based localization techniques for robotic applications

Robotics;Localization;Particle filter;Time of flight

Aranguiz, Dario ; Do ; Minh

关键词: Robotics;    Localization;    Particle filter;    Time of flight;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/99528/ARANGUIZ-THESIS-2017.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

This thesis provides an overview of the localization techniques used by the robotic platform Mousr, as well as the modeling assumptions and simulator scaffolding used to test these techniques. Mousr is a centimeter-scale differential drive robot with an IMU, wheel encoders, and a front-facing time-of-flight sensor. We fuse the differential IMU and wheel encoder measurements with the absolute time-of-flight sensor measurements via particle filter to produce an accurate localization estimate within a known map. Our particle filter successfully accounts for the nonlinearities of the time-of-flight sensor model; we are able to driftlessly drive around a known map in perpetuity. We also discuss a method to approximate the information gain of a map, a crucial component of active localization techniques. Our Jensen-based approximation runs over 20,000 times faster than the exact computation within a 2m x 2m region with minimal degradation in accuracy.