2020/03/25 by V. D. Sharma, Maymoonah Toubeh, Sharma, Vishnu D. +5 · 3 citations
Computer Science · #AI-based Problem Solving and Planning #Anomaly Detection Techniques and Applications #Bayesian Modeling and Causal Inference #FOS: Computer and information sciences #Robotics (cs.RO)
paper · pdf · doi:10.48550/arxiv.2003.11675
openalex publication_date 2020/03/25 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
We propose a risk-aware framework for multi-robot, multi-demand assignment\nand planning in unknown environments. Our motivation is disaster response and\nsearch-and-rescue scenarios where ground vehicles must reach demand locations\nas soon as possible. We consider a setting where the terrain information is\navailable only in the form of an aerial, georeferenced image. Deep learning\ntechniques can be used for semantic segmentation of the aerial image to create\na cost map for safe ground robot navigation. Such segmentation may still be\nnoisy. Hence, we present a joint planning and perception framework that\naccounts for the risk introduced due to noisy perception. Our contributions are\ntwo-fold: (i) we show how to use Bayesian deep learning techniques to extract\nrisk at the perception level; and (ii) use a risk-theoretical measure, CVaR,\nfor risk-aware planning and assignment. The pipeline is theoretically\nestablished, then empirically analyzed through two datasets. We find that\naccounting for risk at both levels produces quantifiably safer paths and\nassignments.\n