Active Point-Cloud Detection for Ghost Object Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Active detection and ranging systems, such as radar, suffer from the issue of detecting 'ghost objects' due to signal reflections from secondary objects, which can lead to incorrect object classification and unsafe actions in the context of ADAS, particularly in advanced driver-assistance systems.
Innovation Solution
A computer-implemented method using a machine learning model with a shared encoder and separate object and ghost classification networks to classify points in a point cloud, employing latent representations and loss functions to improve classification accuracy and distinguish between real and ghost objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single classification network is used to classify objects in the point cloud, then the device complexity is reduced, but the classification precision for distinguishing ghost objects from real objects deteriorates
Solution Approach 1:
The patent divides the classification task into two separate networks: an object classification network that identifies real objects (pedestrian, vehicle, cyclist) and a ghost classification network that detects ghost objects. This segmentation allows each network to specialize in its specific classification task, improving overall detection accuracy while maintaining manageable system complexity through modular architecture.
2Measurement precision
If separate classification networks are used for object and ghost classification, then the classification precision improves, but the device complexity increases
Solution Approach 1:
The patent merges the feature extraction functionality into a shared encoder network that both the object classification network and ghost classification network utilize. This combining approach reduces redundant computations and memory requirements while maintaining the benefits of separate specialized classification networks, thus improving classification accuracy without proportionally increasing device complexity.
3Productivity
If traditional machine learning classification is used, then the processing speed is adequate, but the classification precision for ghost objects deteriorates
Solution Approach 1:
The patent introduces latent representations as an intermediary feature space between the raw point cloud data and the final classification outputs. The shared encoder network transforms input data into latent representations that capture essential features for both object and ghost classification, enabling the system to achieve high precision without sacrificing processing speed by performing computations in this intermediate feature space.
Data Source
AI summary
A computer-implemented method for classifying points in a point cloud obtained by an active detection and ranging system is provided. The method comprises: computing a latent representation of the point cloud, determining from the latent representation, for each point in the point cloud, a probability that the point represents an object belonging to one or more object classes, and determining from the latent representation, for each point in the point cloud, a probability the point represents a ghost object. An object classification network is used for determining the probability that the point represents an object, and a ghost classification network is used for determining the probability the point represents a ghost object.