Learning-based Measurement Scheduling for Loosely-Coupled Cooperative Localization

In cooperative localization (CL), communicating mobile agents use inter-agent relative measurements to improve their dead-reckoning-based global localization. Measurement scheduling enables an agent decide which subset of available it should process when its computational resources are limited. Optimal measurement is NP-hard combinatorial optimization problem. The so-called sequential greedy (SG) algorithm a popular suboptimal polynomial-time solution for this However, the merit function evaluation SG algorithms requires access state estimate vector and error covariance matrix all landmark (teammates that can take from). This paper proposes CL follows approach but reduces communication computation cost by using neural network-based surrogate model as proxy function. significance driven local information only scalar metadata from agents. addresses time memory complexity issues running in three ways: (a) reducing message size, (b) decreasing evaluations simpler (proxy) function, (c) required size. Simulations demonstrate our results.