Optimal Placement of Roadside Infrastructure Sensors towards Safer Autonomous Vehicle Deployments

Abstract

Vehicles with driving automation are increasingly being developed for deployment across the world. However, the onboard sensing and perception capabilities of such automated or autonomous vehicles (AV) may not be sufficient to ensure safety under all scenarios and contexts. Infrastructure-augmented environment perception using roadside infrastructure sensors can be considered as an effective solution, at least for selected regions of interest such as urban road intersections or curved roads that present occlusions to the AV. However, they incur significant costs for procurement, installation and maintenance. Therefore these sensors must be placed strategically and optimally to yield maximum benefits in terms of the overall safety of road users. In this paper, we propose a novel methodology towards obtaining an optimal placement of V2X (Vehicle-to-everything) infrastructure sensors, which is particularly attractive to urban AV deployments, with various considerations including costs, coverage and redundancy. We combine the latest advances made in raycasting and linear optimization literature to deliver a tool for urban city planners, traffic analysis and AV deployment operators. Through experimental evaluation in representative environments, we prove the benefits and practicality of our approach.

Virtual‐real‐fusion simulation framework for evaluating and optimizing small‐spatial‐scale placement of cooperative roadside sensing units

“…Because of the user-defined 3D geometry's inflexibility in modeling diverse scenarios, the majority of 3D studies optimized visual sensors' placement on the simulation platform including Autodesk Revit López et al, 2023), Autodesk CAD (Rougeron et al, 2022), RoadRunner (Vijay et al, 2021), Unity (Jin et al, 2022), Unreal (Hermann et al, 2022), and CARLA (Cai et al, 2023;Qu et al, 2023), as noted in Table 1. The advantages of simulation software are twofold: (1) Users can create customized scenarios by assembling 3D models provided in the software and (2) the built-in ray-casting algorithms Kim & Park, 2019;Chen, Zhu, et al, 2021;Vijay et al, 2021;Hermann et al, 2022;Rougeron et al, 2022;Qu et al, 2023; for rendering make it easier to emulate sensors' data capturing.…”

CoPEM: Cooperative Perception Error Models for Autonomous Driving

“…Identifying the location of RSPUs and their sensors is a major problem in this context, as their optimal placement can drastically improve safety and reduce cost (e.g., procurement, installation, and maintenance). Many studies aim to optimize coverage [9], [10]. To achieve this, some authors consider probabilistic coverage, introducing probabilistic sensor models [11]- [13].…”

“…Moreover, we experimented with RSPU positioning, and simplified the experiment set by considering only optimal placements (refer [10] for details) that actually help overcome the occlusion. Without this step, there could be redundant RSPU placements and coPEM configurations that may result in the same or similar AV behavior as with the regular PEM configurations (non-cooperative perception).…”

Sensor Visibility Estimation: Metrics and Methods for Systematic Performance Evaluation and Improvement

“…Coverage is estimated with raytracing to optimize an infrastructure sensor setup for surveillance by Altahir et al [4] and for vehicle-to-everything (V2X) by Vijay et al [5]. They define coverage as the percentage of sensors that can see any given point on the surface.…”