Evidential Trustworthiness Estimation for Cooperative Perception
Estimation Évidentielle de Confiance pour la Perception Coopérative
Abstract
Intelligent Vehicles can exchange their perception information using wireless technology in a cooperative and decentralized manner. This has the potential to extend the range of perception and thus improve anticipation for complex driving maneuvers and decision making. However, information received from other peers can be erroneous and has to be used carefully. In this paper, we present a method that allows each peer to assign a trust in the information received from other peers based on comparisons with its current knowledge of the world. We describe how this process is managed using the Dempster-Shafer theory. We also present how positive and negative evidence cues can be developed in this problem, in particular by using detectability grids. An experimental evaluation, carried out with real vehicles, is reported to show that this formalism behaves correctly.
Origin : Files produced by the author(s)