Vehicle Object Classification via Height Difference Threshold
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driver assistance systems face challenges in accurately distinguishing between objects that can be driven under, such as gantries or speed indicators, and relevant objects on the roadway, like lost cargo or static obstacles, due to difficulties in sensor data classification.
Innovation Solution
A method that uses sensor data from environmental sensors to determine the height of objects relative to the vehicle and the roadway, classifying objects as drivable if the difference between the object's base point height and the roadway height exceeds a predefined threshold, thereby differentiating between objects that can be driven under and those that require attention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If sensor data is projected into camera image for object classification, then object recognition can be performed using object recognition algorithms, but the complexity of the system increases and measurement precision deteriorates due to coordinate transformation errors
Solution Approach 1:
The patent introduces an intermediary coordinate transformation process that uses a precisely calibrated camera-lidar extrinsic parameter matrix as a mediator to transform lidar point cloud data into camera image coordinates. This intermediary transformation maintains measurement precision by using pre-calibrated transformation parameters rather than direct projection, thereby resolving the contradiction between ease of operation and measurement precision.
2Reliability
If probability models are used to determine stationary vehicles and upper objects, then classification can be performed, but the device complexity increases due to multiple correlation models and threshold determinations
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing extrinsic parameter matrices and transformation relationships between sensor coordinate systems before actual object detection. These pre-computed transformation parameters are then reused during runtime to simplify the classification process, reducing device complexity while maintaining classification reliability through accurate coordinate transformations.
3Measurement precision
If lateral lidar sensors are arranged and tilted to detect upper spatial regions, then detection capability for bridges and tunnel entrances is improved, but the device complexity and installation complexity increase
Solution Approach 1:
The patent makes the lidar sensor system universal by configuring it to perform multiple functions: detecting both ground-level objects and upper spatial regions such as bridges and tunnel entrances. The lidar sensor can operate in different modes (ground scanning and upper region scanning) without requiring separate dedicated sensors, thereby improving measurement precision for upper regions while reducing device complexity through multi-functionality.
Data Source
AI summary
A method for classifying objects includes receiving sensor data from an environment sensor of the vehicle; recognizing, based on the sensor data, an object in a region of a roadway; determining an object region on the roadway with which the object is associated; assigning a base point to the object based on the sensor data; determining the height of the base point with respect to a vehicle vertical direction; determining the roadway height with respect to the vehicle vertical direction in the object region under the assumption of a predetermined grade of the roadway between a forward-zone region of the roadway in front of the vehicle and the object region; and classifying the object as an object that can be driven under, if the difference between the roadway height and the height of the base point exceeds a predefined threshold value.