Vehicle Feature Detection via Abstract Sensor Space Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle assistance systems face challenges in efficiently determining features in high-dimensional sensor data from environments, leading to high computational loads and loss of information, especially when using radar sensors for semi-autonomous driving.
Innovation Solution
A computer-implemented method that maps measurement points from high-dimensional data spaces into abstract signal processing spaces with reduced dimensionality, allowing for efficient feature detection by preserving topological information and reducing data processing power, using techniques like self-organizing maps and multilayer perceptrons to optimize the feature detection process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-dimensional sensor data is processed directly to detect features, then detection accuracy is improved, but computational load increases significantly
Solution Approach 1:
The patent transforms the high-dimensional sensor data from radar (range, azimuth, Doppler velocity) into a two-dimensional time-distance image by introducing the time dimension. This dimensionality transformation allows the system to process complex 3D spatial data as 2D images, reducing computational complexity while preserving essential feature information for detection.
Solution Approach 2:
The patent introduces an intermediary representation called 'time-distance image' that mediates between the raw high-dimensional radar data and the final feature detection. This intermediate form consolidates multiple dimensions into a 2D representation that is computationally more efficient to process while retaining the necessary information for accurate feature detection.
2Productivity
If data dimensionality is reduced to decrease computational load, then processing efficiency is improved, but information loss occurs
Solution Approach 1:
Instead of reducing dimensions by aggregation or projection, the patent transforms the data into a different dimensional representation where the time dimension becomes the vertical axis and range becomes the horizontal axis. This transformation preserves contextual information by maintaining the temporal evolution of targets, allowing efficient 2D processing without significant information loss.
3Reliability
If contextual information is incorporated to improve detection reliability, then detection reliability is improved, but system complexity increases
Solution Approach 1:
The patent incorporates contextual information by utilizing the time dimension to show the evolution and movement of targets over time. This temporal context is integrated naturally into the 2D time-distance image representation, allowing reliable feature detection without requiring complex multi-sensor fusion or sophisticated algorithms. The temporal context provides reliability while the 2D representation maintains simplicity.
Data Source
Figure 1a
Figure 1b
Figure 2a
AI summary
The present disclosure relates to a computer-implemented method, apparatus, computer program and a vehicle comprising the apparatus for determining one or more features of a surrounding of a vehicle, the method comprising the steps of: obtaining sensor data comprising measurement points, the measurement points being arranged in a data space; determining an abstract signal processing space based on a distribution of the measurement points in the data space; mapping the measurement points into the abstract signal processing space; and determining the one or more features in the abstract signal processing space based on an arrangement of the measurement points in the abstract signal processing space.