Active perception for autonomous vehicles

Abstract

Precise perception of a vehicle's surrounding is crucial for safe autonomous driving. It requires a high sensor resolution and a large field of view. Active perception, i.e. the redirection of a sensor's focus of attention, is an approach to provide both. With active perception, however, the selection of an appropriate sensor orientation becomes necessary. This paper presents a method for determining the sensor orientation in urban traffic scenarios based on three criteria: the importance of traffic participants w.r.t. the current situation, the available information about traffic participants while considering alternative sensor orientations as well as sensor coverage of the vehicle's relevant surrounding area.

Safe Occlusion-Aware Autonomous Driving via Game-Theoretic Active Perception

“…The idea of active perception [22][23][24] attempts to close the loop between robot perception and motion planning by generating trajectories that will explore the environment and gain information to reduce perception uncertainty and improve future motion planning. Some recent works [25,26] also tackle the occlusion-aware trajectory planning problem through active perception by incorporating the prediction of the autonomous vehicle's future field-of-view into motion planning.…”

Situation-Aware Environment Perception Using a Multi-Layer Attention Map

“…Before this increase, research in the field of active perception employed actuated sensors or sensor platforms to actively control gazing direction, e.g. [8], [16]- [18]. Defining the gazing direction over time is commonly represented by an optimization function that needs to be maximized.…”

Semi-task-dependent and uncertainty-driven world model maintenance

“…The work of Unterholzner et al (2012) introduces active perception in an autonomous driving setting. A utility function combines object importance with the expected reduction of the uncertainty of a world state estimate after detecting an object.…”