Radar Object Identification Using LiDAR Ground-Truth Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating the quality of object identification by radar devices in motor vehicles are inefficient and costly, lacking objective criteria for optimization.
Innovation Solution
A method involving LiDAR scanning to generate high-resolution maps, followed by object identification and creation of ground-truth models, with subsequent comparison to radar device maps using alternative algorithms to determine quality functions and select the most suitable algorithm, aided by generative statistical models for simulating radar data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional methods are used to evaluate radar device object identification quality, then development can proceed with existing processes, but the determination of quality is time-consuming and costly
Solution Approach 1:
The patent creates virtual copies of radar measurements through simulation models that replicate real sensor behavior. These virtual measurements are generated from digital twin models of the environment and radar system, allowing quality evaluation without physical testing. The simulation models include radar signal generation, propagation, and reception characteristics, enabling accurate quality assessment in virtual environments.
Solution Approach 2:
The patent performs quality determination activities in advance during the development phase using simulated data. By evaluating object identification quality before actual deployment or physical testing, the system identifies and corrects issues early in the development process, reducing overall development time and costs while maintaining reliability.
2Measurement precision
If multiple alternative object identification algorithms are tested to find the most suitable one, then object identification quality can be optimized, but the evaluation process becomes more complex
Solution Approach 1:
The determination apparatus automatically evaluates multiple object identification algorithms using simulated radar measurements and objectively determines their quality without manual intervention. The system self-manages the comparison process, algorithm testing, and quality ranking, reducing the complexity burden on developers while maintaining precise object identification assessment.
Solution Approach 2:
The patent systematically varies parameters of different object identification algorithms and evaluates their performance using simulated radar data. By changing algorithm parameters such as detection thresholds, clustering methods, and filtering settings, the system objectively identifies the optimal configuration for accurate object identification while managing complexity through automated parameter sweeps.
3Reliability
If comprehensive quality certification is performed according to legal regulations, then radar device reliability is ensured, but development costs increase
Solution Approach 1:
The patent uses virtual copies and simulation models to perform comprehensive quality certification activities that would otherwise require expensive physical testing and validation. The digital twin environment replicates real-world conditions and regulatory test scenarios, enabling cost-effective quality assurance while maintaining the rigor required by legal regulations.
Solution Approach 2:
The system performs quality certification activities in advance during development using simulated measurements and data. By completing comprehensive quality assessments before production or deployment, the system identifies and resolves issues early, avoiding costly rework or recalls later while ensuring regulatory compliance is met during the development phase.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables automated, cost-effective determination of object identification quality, reducing development times and costs by identifying reliable object identification algorithms for radar devices.
Implementation Method 1
generating a first map of the surroundings using a LiDAR scanner
Implementation Method 2
generating a first map of the surroundings using a LiDAR scanner
Implementation Method 3
generating a second map of the surroundings by means of the radar device
Data Source
AI summary
A method for improving an evaluation of object identification of a radar device of a motor vehicle. A first map is generated of the surroundings of said surroundings by a LiDAR scanner. A first object identification is performed on the basis of the generated first map of the surroundings by a first identification apparatus. A ground-truth model is created of the surroundings on the basis of the performed first object identification by the first identification apparatus. A second map is generated of the surroundings of the surroundings by the radar device. A second object identification is repeatedly performed using alternative object identification algorithms on the basis of the generated second map of the surroundings by a second identification apparatus of the radar device. A radar model is created of the surroundings on the basis of the repeatedly performed second object identification by means of the second identification apparatus.