Ultrasonic Obstacle Recognition Using Occupancy Grid Risk Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ultrasonic sensors have limitations in accurately estimating the shape and location of obstacles, leading to incorrect braking, especially when encountering curbs or narrow passages, due to expanded contour recognition and failure to distinguish passable obstacles.
Innovation Solution
An obstacle recognition device utilizing a learning algorithm with a recurrent neural network (RNN) model, processing direct and indirect waves, signal strength, and vehicle behavior to generate an occupancy grid map, calculate risk probabilities, and determine obstacle shape and location, thereby improving collision-avoidance assist control and minimizing incorrect braking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional ultrasonic sensor is used to measure obstacle location and shape, then distance information can be obtained, but the obstacle contour is expanded and shape estimation is inaccurate
Solution Approach 1:
The patent divides the obstacle detection space into multiple scanning lines and segments the ultrasonic sensor array into multiple sensing units. Each sensing unit detects obstacles along specific scanning lines, allowing the system to reconstruct the true obstacle contour by integrating data from multiple segments rather than treating the entire sensor array as a single unit.
Solution Approach 2:
The patent introduces a new dimension of analysis by considering the spatial distribution of ultrasonic signals across multiple scanning lines and sensor units. Instead of relying on a single distance measurement, the system analyzes the pattern of signal reflections across multiple dimensions (different sensor units, different scanning lines, different time delays) to reconstruct the true obstacle shape and location.
2Reliability
If conventional ultrasonic sensor recognizes obstacle based on distance information, then obstacle detection is performed, but passable obstacles like curbs cannot be distinguished leading to incorrect braking
Solution Approach 1:
The patent applies local quality analysis by examining the reflection characteristics of ultrasonic signals at different locations on the obstacle surface. By analyzing which specific sensor units receive reflected signals and from which scanning lines, the system can determine local geometric features of the obstacle, such as whether it has a vertical face (non-passable) or a sloped surface (passable).
Solution Approach 2:
The patent introduces an intermediate processing layer that analyzes the pattern of ultrasonic signal reflections before making collision avoidance decisions. This intermediary analysis of signal characteristics and reflection patterns provides additional information about obstacle type, enabling the system to distinguish between passable and non-passable obstacles before triggering braking control.
3Measurement precision
If multiple sensing information of ultrasonic sensor, camera, and radar are used, then obstacle location and shape can be measured, but the complexity of the system increases
Solution Approach 1:
The patent makes the ultrasonic sensor array multi-functional by enabling it to perform both distance measurement and shape recognition tasks using the same hardware. By processing ultrasonic signal reflections through multiple scanning lines and analyzing patterns across different sensor units, the system achieves both functions with a single sensor type, eliminating the need for separate camera and radar systems.
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
The solution accurately estimates obstacle shapes and locations, preventing incorrect braking by distinguishing passable obstacles and calculating appropriate braking control times based on risk probabilities and vehicle speed, enhancing low-speed driving and parking control accuracy.
Implementation Method 1
distance information obtained via an ultrasonic sensor
Implementation Method 2
sequential direct and indirect waves and a signal strength from the at least one ultrasonic sensor
Data Source
AI summary
An obstacle recognition device, a vehicle system including the same, and a method thereof are provided. The obstacle recognition device includes a storage storing data and an algorithm for calculating a risk probability and a processor configured to execute the algorithm to generate an occupancy grid map based on a sensing value of at least one ultrasonic sensor, calculate the risk probability of each cell on the occupancy grid map, and determine a shape and location of an obstacle based on the risk probability of each cell.