Autonomous Occupancy Detection Through Neural Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection and map generation methods for autonomous systems rely heavily on single sensor modalities like LiDAR, RADAR, or ultrasonic sensors, which are prone to noise, errors, and high computational costs, leading to unreliable object maps, especially at varying distances and environmental conditions.
Innovation Solution
A neural network-based approach that fuses data from multiple sensor modalities, including ultrasonic, image, and RADAR sensors, to generate more reliable object maps by extracting and combining feature datasets, thereby leveraging the strengths of each modality while mitigating their weaknesses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR is used to generate object maps, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent combines data from multiple sensor modalities (ultrasonic, RADAR, and image sensors) into a unified object map. This fusion approach merges the advantages of each sensor type to achieve high measurement precision without relying solely on expensive LiDAR systems, thereby reducing device complexity while maintaining detection accuracy.
Solution Approach 2:
The system employs a multi-functional sensor suite where ultrasonic sensors handle close-range detection, RADAR provides mid-range detection, and image sensors capture visual information. This multi-functional approach replaces the need for expensive LiDAR by distributing detection functions across multiple cheaper sensor types, reducing overall system complexity while maintaining precision.
2Device complexity
If RADAR data is used for object detection, then device complexity is reduced, but measurement precision deteriorates at close distances
Solution Approach 1:
The patent applies local quality by assigning different sensor types to different spatial zones: ultrasonic sensors are used for close-range detection where RADAR precision deteriorates, while RADAR handles mid-range detection. This spatial differentiation of sensor functions maintains high measurement precision across all distances while keeping the system relatively simple.
Solution Approach 2:
The system merges RADAR data with ultrasonic sensor data to compensate for RADAR's weaknesses at close distances. By combining these sensor modalities, the system achieves reliable detection across all distance ranges without requiring complex LiDAR systems, thus maintaining low device complexity while improving overall measurement precision.
3Device complexity
If ultrasonic sensors are used for object detection, then device complexity is reduced, but measurement precision deteriorates at distances greater than three meters
Solution Approach 1:
The system applies local quality by using ultrasonic sensors for close-range detection (within three meters) where they excel, and transitioning to RADAR for mid-range detection beyond three meters. This spatially differentiated sensor deployment maintains high measurement precision across all distances while keeping the system simple and cost-effective.
Solution Approach 2:
The patent merges ultrasonic sensor data with RADAR data to create a seamless detection system that leverages the strengths of each sensor type at different distance ranges. This combination allows the system to maintain high measurement precision from close to far distances without requiring complex LiDAR systems, thus reducing device complexity while improving overall detection accuracy.
4Device complexity
If image sensors are used for object detection, then device complexity is reduced, but reliability deteriorates in poor lighting and weather conditions
Solution Approach 1:
The system merges image sensor data with RADAR and ultrasonic sensor data to compensate for image sensors' weaknesses in poor lighting and weather conditions. By fusing data from these multiple modalities, the system maintains high detection reliability across all environmental conditions while keeping device complexity relatively low compared to LiDAR-based systems.
Solution Approach 2:
The patent creates a composite sensing system that combines data from multiple sensor modalities (ultrasonic, RADAR, and image sensors) to detect objects. This composite approach is analogous to using composite materials, where each sensor type contributes its strengths to overcome the weaknesses of individual sensors, thereby maintaining high reliability in varying environmental conditions while keeping the system simple.
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
Improves the reliability and accuracy of object maps by integrating data from multiple sensors, reducing computational demands and overcoming limitations of single-sensor approaches.
Implementation Method 1
ultrasonic sensor data corresponding to an area may be obtained
Implementation Method 2
RADAR data corresponding to the area may be obtained
Data Source
AI summary
The present disclosure relates to performing sensor and/or temporal fusion for occupancy determinations in autonomous or semi-autonomous systems and applications. For example, ultrasonic data, image data, and RADAR data may be processed using one or more neural networks to generate output data corresponding to one or more objects in an area. During processing, a first feature dataset, a second feature dataset, and a third feature dataset may be extracted from the ultrasonic sensor data, the image data, and the RADAR data, respectively, and a combined feature dataset corresponding to the output data may be generated based at least on the first feature dataset, the second feature dataset and the third feature dataset. A machine may be caused to perform one or more operations based at least on the output data.