Visual-Guided LiDAR Scanning for Key-Target Trajectory Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-driving vehicle recognition systems face challenges such as long evolution cycles, data leakage risks, mechanical damage, limited detection range, high costs, and inefficient scanning in pure visual and laser radar systems, and millimeter-wave radar systems are only suitable for obstacle detection.
Innovation Solution
A laser radar driving environmental recognition system combining visual and laser radar systems to accurately measure key targets and predict movement trajectories, using a visual system, laser radar system, data processing module, streaming media analysis, key target spherical coordinate data processing, and three-dimensional trajectory prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If pure visual recognition system is used to achieve wide-area recognition capability, then recognition coverage is improved, but evolution cycle becomes long and data leakage risks increase
Solution Approach 1:
The patent combines visual recognition system and laser radar system into a hybrid recognition system. The visual system performs wide-area screening to identify key target regions, while the laser radar system conducts precise measurement only in these key areas. This merging allows the system to achieve comprehensive coverage with reduced computational burden, avoiding the long evolution cycle of pure visual systems.
2Measurement precision
If laser radar recognition system increases laser spot scanning density to improve recognition accuracy, then measurement precision is improved, but scanning resources are wasted and scanning range becomes excessively large
Solution Approach 1:
The patent applies different scanning densities to different spatial regions. The visual system identifies key target areas that require high scanning density for accurate measurement, while other regions use lower scanning density or are excluded from laser scanning entirely. This local differentiation of scanning quality achieves high recognition accuracy for critical targets while minimizing overall scanning resource consumption.
Solution Approach 2:
The patent segments the scanning area into key target regions identified by the visual system and non-key regions. The laser radar system focuses its high-density scanning only on the segmented key regions, rather than uniformly scanning the entire field of view. This segmentation allows precise measurement where needed while avoiding unnecessary scanning elsewhere, reducing resource waste.
3Measurement precision
If existing laser radar recognition system continuously increases number of measuring points to improve recognition accuracy, then measurement precision is improved, but detection cost increases
Solution Approach 1:
The patent performs preliminary action by using the visual recognition system to pre-identify key target areas and objects before the laser radar system conducts detailed measurement. This preliminary screening determines which regions require high-precision laser measurement, allowing the system to increase measuring points only where necessary for accuracy while reducing the overall number of measurements required, thereby lowering detection cost.
4Area of stationary object
If millimeter-wave radar recognition system is used for obstacle detection, then detection range is improved, but detection accuracy becomes limited
Solution Approach 1:
The patent merges millimeter-wave radar, visual recognition, and laser radar systems into a multi-sensor hybrid system. The millimeter-wave radar provides wide-area obstacle detection capability, while the visual system identifies key targets, and the laser radar delivers precise measurements. This combination allows the system to achieve both extensive detection range and high measurement precision that no single system could achieve alone.
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
Achieves accurate measurement and prediction of key targets, avoiding excessive calculation and inefficient scanning, while enhancing detection capabilities and reducing resource waste.
Implementation Method 1
a key target spherical coordinate data processing module, configured for analyzing and processing spherical coordinate data of key targets output by the streaming media analysis module, and outputting azimuth coordinates in the processed spherical coordinates of key targets to the laser radar system; a key target three-dimensional trajectory prediction module, wherein the laser radar system is configured for measuring real distances of the key targets
Data Source
AI summary
Embodiments of the present disclosure provide a laser radar driving environmental recognition system based on visual area guidance, including a visual system, a laser radar system, a data processing module and a vehicle action control module. The data processing module includes: a streaming media analysis module; a key target spherical coordinate data processing module; a key target three-dimensional trajectory prediction module, wherein the laser radar system is configured for measuring real distances of the key targets, and outputting real spherical coordinates of key targets to the key target three-dimensional trajectory prediction module, the key target three-dimensional trajectory prediction module being configured for generating real-time trajectory prediction data of key targets, and outputting the real-time trajectory prediction data of key targets to the vehicle action control module. The laser radar driving environmental recognition system can achieve accurate measurement of key targets and prediction of movement trajectories.