The Radar Ghost Dataset – An Evaluation of Ghost Objects in Automotive Radar Data

Abstract

Radar sensors have a long tradition in advanced driver assistance systems (ADAS) and also play a major role in current concepts for autonomous vehicles. Their importance is reasoned by their high robustness against meteorological effects, such as rain, snow, or fog, and the radar's ability to measure relative radial velocity differences via the Doppler effect. The cause for these advantages, namely the large wavelength, is also one of the drawbacks of radar sensors. Compared to camera or lidar sensor, a lot more surfaces in a typical traffic scenario appear flat relative to the radar's emitted signal. This results in multi-path reflections or so called ghost detections in the radar signal. Ghost objects pose a major source for potential false positive detections in a vehicle's perception pipeline. Therefore, it is important to be able to segregate multipath reflections from direct ones. In this article, we present a dataset with detailed manual annotations for different kinds of ghost detections. Moreover, two different approaches for identifying these kinds of objects are evaluated. We hope that our dataset encourages more researchers to engage in the fields of multi-path object suppression or exploitation.

A noise analysis of 4D RADAR: robust sensing for automotive?

“…Given the importance of the quality of the data provided by the senor suite, the number of recent works around perception sensors and the effect of weather conditions has increased dramatically [16]- [26]. Moreover, some novel works have proposed to explore and analyse in a holistic way the noise factors which camera and LiDAR are exposed to in the automotive environment, studying the effect of single and compound noise factors [27]- [29].…”

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

“…It provides three kinds of radar data, including the RAD tensor, RA map, and BEV map. The Radar Ghost dataset [61] aims at studying the effect of multi-path propagation in autonomous driving. It provides pointwise annotations of real targets and four types of ghost targets.…”

“…Therefore, using LiDAR as the ground truth could be sometimes problematic. In the Radar Ghost dataset [61], ghost objects are manually annotated with the help of a helper tool. This tool can automatically calculate the locations of potential ghosts based on real objects and reflective surfaces.…”

“…Take ghost detection as an example. From a data perspective, we need to use our expert knowledge to label ghost objects [61]. From a network perspective, we can design an attention module [201] or utilise graph convolution [167] to model the relationship between real and ghost objects.…”

Misspecified Cramér-Rao Bound for Multipath Model in MIMO Radar

“…Elevation multipath may also occur in tunnels, below bridges, or over-path road signs and constructions [30]. Horizontal multipath occurs when driving near guardrails, buildings, and adjacent vehicles [31,32].…”