Online Mapping and Perception Algorithms for Multi-robot Teams Operating in Urban Environments.

field robotics;mapping;segmentation;camera calibration;signal-strength prediction;occupancy-grid;Computer Science;Engineering;Computer Science and Engineering

Strom, Johannes H.Hsieh, M. Ani ;

University of Michigan

关键词: field robotics;    mapping;    segmentation;    camera calibration;    signal-strength prediction;    occupancy-grid;    Computer Science;    Engineering;    Computer Science and Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/113516/jhstrom_1.pdf?sequence=1&isAllowed=y

瑞士|英语

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

【 摘 要 】

This thesis investigates some of the sensing and perception challenges facedby multi-robot teams equipped with LIDAR and camerasensors. Multi-robot teams are ideal for deployment in large,real-world environments due to their ability to parallelize exploration,reconnaissance or mapping tasks.However, such domains also impose additional requirements, including theneed for a) online algorithms (to eliminate stopping and waiting forprocessing to finish before proceeding) and b) scalability (to handledata from many robots distributed over a large area).These general requirements give rise to specific algorithmic challenges, including 1) online maintenance of large, coherentmaps covering the explored area, 2) online estimation of communication propertiesin the presence of buildings and other interfering structure, and 3)online fusion and segmentation of multiple sensors to aid in object detection.The contribution of this thesis is the introduction of novelapproaches that leverage grid-maps and sparse multi-variate gaussianinference to augment the capability of multi-robot teams operating inurban, indoor-outdoor environments by improving the state of the artof map rasterization, signal strength prediction, colored point cloudsegmentation, and reliable camera calibration.In particular, we introduce a map rasterization technique for largeLIDAR-based occupancy grids that makes online updates possible whendata is arriving from many robots at once.We also introduce newonline techniques for robots to predict the signal strength to theirteammates by combining LIDAR measurements with signal strengthmeasurements from their radios.Processing fused LIDAR+camera pointclouds is also important for many object-detection pipelines. Wedemonstrate a near linear-time online segmentation algorithm to thisdomain. However, maintaining the calibration of a fleet of 14 robotsmade this approach difficult to employ in practice.Therefore we introduced a robust and repeatablecamera calibration process that grounds the camera model uncertainty in pixelerror, allowing the system to guide novices and experts alike to reliably produce accurate calibrations.