Test system and detection method for target identification and monitoring of monocular camera of intelligent vehicle

Abstract

The invention discloses an intelligent vehicle monocular camera target identification monitoring test system and a detection method. The system comprises a main body experiment platform, a visual scene generation simulation subsystem, a monocular camera three-degree-of-freedom motion platform, a rain and fog weather environment generation subsystem, a main control computer experiment management subsystem, a target identification test subsystem and an identification information acquisition subsystem. The main body experiment platform is connected with a monocular camera three-degree-of-freedom motion platform, a scene generation simulation subsystem and a rain and fog weather environment generation subsystem, the monocular camera three-degree-of-freedom motion platform is connected with an identification information acquisition system, and the identification information acquisition system is connected with a main control computer test management subsystem. The invention has the beneficial effects that the recognition evaluation subsystem is developed by comparing standard data, fusing the accuracy rate, the recall rate and the real-time index, and a test method process is designed for comprehensively evaluating the target recognition effect, so the defects of low randomness and objectivity of manual evaluation are overcome.

Description

technical field [0001] The invention relates to the field of smart cars, in particular to a target recognition monitoring test system and a detection method of a monocular camera of a smart car. Background technique [0002] Since a traffic accident is a low-probability event, the probability of a fatal accident in a traffic accident is about 1.5×10-9 / km, so it requires a very high workload to intuitively prove the safety of smart cars from a statistical point of view. At a 95% confidence level, approximately 8 billion kilometers of actual road testing are required to prove that smart cars cause lower traffic accident fatalities than human drivers. That is, assuming that a test fleet composed of 100 vehicles is tested 24 hours a day, 365 days a year, at an average speed of 40km / h, it will take 225 years to complete this mileage. At the same time, during the testing process of smart cars, it also exposed the shortcomings of smart cars in environment perception and decision-m...

AI Technical Summary

Problems solved by technology

It is necessary to reproduce typical traffic hazard scenarios in a dedicated controlled site, involving different road shapes, different lane alignments, different traffic signs, different traffic participants (such as vehicles, pedestrians, cyclists), etc., the complexity High, highly specialized, high site operation and management costs, high test loss, empirical tests, because the probability of dangerous conditions during vehicle driving is very small, a large number of vehicle tests are required to obtain statistically significant results, resulting in test data The amount of traffic is very large, and each empirical test requires a lot of manpower, material resources and funds, and there is a possibility of real traffic accidents causing casualties

Images