RF Signal Analysis for ADAS Object Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current advanced driver assistance systems (ADAS) face limitations in detecting and classifying non-vehicular road users, such as pedestrians and cyclists, due to visual obstructions and reliance on sensors that require a clear line of sight, leading to reduced detection accuracy and increased false alarms.

Innovation Solution

A system that utilizes wireless transmission analysis from devices associated with targets, such as smartphones and Bluetooth devices, to detect and classify objects within a physical environment, enhancing detection capabilities by integrating with existing ADAS systems.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If sensors (cameras, radar, LiDAR) are used for object detection in ADAS systems, then detection capability is improved, but detection accuracy deteriorates when visual obstructions are present

Engineering Contradiction:
Improvedetection capabilityVSAvoiddetection accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent introduces wireless communication signals as an intermediary detection modality. Instead of relying solely on visual sensors that require line-of-sight, the system uses RF signals (cellular, Wi-Fi, Bluetooth) emitted by devices carried by road users as a mediator to detect and classify objects obscured from visual view. This complementary approach allows detection through obstructions while maintaining accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If wireless transmission analysis is added to enhance detection of obscured objects, then detection accuracy for non-vehicular road users is improved, but device complexity increases

Engineering Contradiction:
Improvedetection accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent leverages the universality of wireless communication infrastructure. The same RF signals used for general communication (cellular networks, Wi-Fi, Bluetooth) are repurposed for detection and classification functions. This multi-functionality approach avoids adding dedicated detection hardware, thereby improving detection accuracy without proportionally increasing system complexity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Reliability

If multiple sensor types are integrated for comprehensive detection, then detection coverage is improved, but false alarms increase

Engineering Contradiction:
Improvedetection coverageVSAvoidfalse alarms
Core Design Contradiction:
ReliabilityVSObject-generated harmful factors

Solution Approach 1:

The patent implements feedback through cross-validation of detection data from multiple modalities. Wireless communication signal characteristics (presence, signal strength, device type) provide feedback that confirms or refutes detections from visual sensors. This feedback mechanism reduces false alarms by requiring corroboration from independent detection channels before confirming obscured object presence.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3980982B1Wireless communication-based classification of objects
Publication Date: 2025.09.03 A D KNIGHT LTD
  • EP3980982B1 patent drawingFigure 1
  • EP3980982B1 patent drawingFigure 2
  • EP3980982B1 patent drawingFigure 3A

AI summary

A method comprising receiving a dataset comprising data associated with a plurality of radio frequency (RF) wireless transmissions associated with a plurality of objects within a plurality of physical scenes, wherein the dataset comprises, with respect to each of the objects, at least: (i) signal parameters of the associated wireless transmissions, (ii) data included in the associated wireless transmissions, and (iii) locational parameters with respect to the object; at a training stage, training a machine learning model on a training set comprising the dataset and labels indicating a type of each of said objects; and at an inference stage, applying the trained machine learning model to a target dataset comprising signal parameters, data, and locational parameters obtained from wireless transmissions associated with a target object within a physical scene, to predict a type of the target object.