CoPEM: Cooperative Perception Error Models for Autonomous Driving

Abstract

In this paper, we introduce the notion of Cooperative Perception Error Models (coPEMs) towards achieving an effective and efficient integration of V2X solutions within a virtual test environment. We focus our analysis on the occlusion problem in the (onboard) perception of Autonomous Vehicles (AV), which can manifest as misdetection errors on the occluded objects. Cooperative perception (CP) solutions based on Vehicle-to-Everything (V2X) communications aim to avoid such issues by cooperatively leveraging additional points of view for the world around the AV. This approach usually requires many sensors, mainly cameras and LiDARs, to be deployed simultaneously in the environment either as part of the road infrastructure or on other traffic vehicles. However, implementing a large number of sensor models in a virtual simulation pipeline is often prohibitively computationally expensive. Therefore, in this paper, we rely on extending Perception Error Models (PEMs) to efficiently implement such cooperative perception solutions along with the errors and uncertainties associated with them. We demonstrate the approach by comparing the safety achievable by an AV challenged with a traffic scenario where occlusion is the primary cause of a potential collision.

Automated Testing With Temporal Logic Specifications for Robotic Controllers Using Adaptive Experiment Design

“…Approaches which assume noise follows a wellknown (e.g., Gaussian) state-independent distribution [7] are insufficient to capture the perceptual variety of a typical AVsystem-a LiDAR detector may be great for close range traffic, but terrible for long range or occluded traffic [36]. We therefore use a Perception Error Model (PEM) [37]-a surrogate trained on real sensor data which mimics perceptual errors encountered in regular operation (Sec II-A).…”

“…In this view, f is composed of a d ′ × d projection matrix H on state x t , plus stochastic error dependent on the current world state w t . The ϵ function is a surrogate model known as a Perception Error Model [37]. This is a probabilistic model of the original AV's perception noise, dependent on salient features g(w) extractable from simulated w. Salient features can include obstacle positions, dimensions, occlusion, or environment factors.…”

Multi-Layered Local Dynamic Map for a Connected and Automated In-Vehicle System

“…Significant contributions to CP research include the following: Seong-Woo Kim et al [16][17][18], who created a framework to extend perception beyond line-of-sight, a cooperative driving system using CP, and methods for improving AD safety and smoothness; Pierre Merdiganc et al [19], who integrated perception and vehicle-to-pedestrian communication to enhance Vulnerable Road Users' (VRUs) safety; Aaron Miller et al [20], who developed a perception and localization system allowing vehicles with basic sensors to leverage data from those with advanced sensors, thus elevating AD capabilities; Xiaboo Chen et al [21,22], who proposed a recursive Bayesian framework for more reliable cooperative tracking, and a robust framework for multi-vehicle tracking under inaccurate self-localization; Adamey et al [23], who introduced a method for collaborative vehicle tracking in mixed-traffic settings; Francesco Biral et al [24], who demonstrated how the SAFE STRIP EU project technology aids in deploying the LDM for Cooperative ITS safety applications; and Stefano Masi et al [25], who developed a cooperative roadside vision system to enhance the perception capabilities of an AV; Sumbal Malik et al [26], who highlight the need for advanced CP to overcome challenges in achieving level 5 AD; Tania Cerquitelli et al [27], who discussed in a special issue the integration of machine learning and artificial intelligence technologies to empower network communication, analysing how computer networks can become smarter; Andrea Piazzoni et al [28], who discuss how to model CP errors in AD, focusing on the impact of occlusion on safety and how CP may address it; Zhiying Song et al [29], who presented a framework for evaluating CP in connected AVs, emphasizing the importance of CP in increasing vehicle awareness beyond sensor FoV; Mao Shan et al [30], who introduced a novel framework for enhancing CP in Connected AVs by probabilistically fusing V2X data, improving perception range and decision-making in complex environments.…”