Automatic lane change scenario extraction and generation of scenarios in OpenX format from real-world data

Abstract

Autonomous Vehicles (AV)'s wide-scale deployment appears imminent despite many safety challenges yet to be resolved. The modern autonomous vehicles will undoubtedly include machine learning and probabilistic techniques that add significant complexity to the traditional verification and validation methods. Road testing is essential before the deployment, but scenarios are repeatable, and it's hard to collect challenging events. Exploring numerous, diverse and crucial scenarios is a time-consuming and expensive approach. The research community and industry have widely accepted scenario-based testing in the last few years. As it is focused directly on the relevant critical road situations, it can reduce the effort required in testing. The scenario-based testing in simulation requires the realistic behaviour of the traffic participants to assess the System Under Test (SUT). It is essential to capture the scenarios from the real world to encode the behaviour of actual traffic participants. This paper proposes a novel scenario extraction method to capture the lane change scenarios using point-cloud data and object tracking information. This method enables fully automatic scenario extraction compared to similar approaches in this area. The generated scenarios are represented in OpenX format to reuse them in the SUT evaluation easily.

Divide and Conquer: A Systematic Approach for Industrial Scale High-Definition OpenDRIVE Generation from Sparse Point Clouds

“…A specialized approach for the extraction of traffic scenarios is shown by Karunakaran et al [16]. The data used in this study was obtained from a vehicle equipped with six LiDAR sensors and an additional ground-facing LiDAR.…”

“…Although these methods achieve a standardized map format, the reliance on special sensor characteristics [16] or mounting positions [15] hinder the adaption of these methodologies. Moreover, generating OpenDRIVE geometry as one consecutive road, as presented in our previous work [9], can lead to errors in large-scale road reconstruction.…”

Automatic Odometry-Less OpenDRIVE Generation From Sparse Point Clouds

“…A higher focus on traffic behaviour is shown in [15]. In addition to a road model, the authors are able to derive and analyse traffic participants.…”

“…Through the nature of point clouds, methods like nearest neighbor would yield wrong connections between lanes. Hence, for the construction of the lanes, related research, e.g., [14], [15], integrates ego trajectory information to solve this limitation, which in return requires an accurate GNSS and/or IMU setup.…”

“…In the case of additional drives, there is no guarantee that the course of the road is unchanged. Furthermore, recent research [12]- [15] has shown to be reliant on specific sensor types, e.g., line-wise scanners, sensor characteristics, like 360°field of view (FoV), or mounting points, which makes the adoption on other data sources complex.…”