Sensor Fusion Object Recognition for Vehicle Collision Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition systems for movable bodies, such as vehicles, struggle to accurately detect and correct errors in recognizing the position and shape of objects outside the vehicle using imaging devices like cameras and radar, leading to potential collisions.
Innovation Solution
A system combining a camera and a millimeter-wave radar to enhance object recognition by using machine learning for image extraction, distance measurement, and correction processing to accurately determine the size and position of objects, including first and second type corrections for different object types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If object recognition is performed using imaging devices like cameras and radar, then object detection capability is improved, but measurement precision of object position and shape deteriorates due to recognition errors
Solution Approach 1:
The system uses radar to measure the actual distance to the object and feeds this information back to correct the position and shape recognition results from the camera. The recognition unit adjusts the detected object parameters based on the radar distance measurement, forming a feedback loop that improves measurement precision while maintaining detection capability.
Solution Approach 2:
The radar acts as an intermediary device that provides distance information to mediate between the camera's image data and the final object recognition results. By introducing this intermediate measurement, the system resolves the contradiction between detection capability and measurement precision.
2Measurement precision
If multiple sensors are combined for object recognition, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The radar serves multiple functions: it provides distance measurement for position correction, scale information for shape correction, and object presence detection. By making the radar multi-functional, the system improves measurement precision without proportionally increasing device complexity.
Solution Approach 2:
The system merges the data from the camera and radar into a unified object recognition process. The recognition unit combines image data with distance measurements to simultaneously determine position, shape, and size of objects, reducing overall system complexity through integration.
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 system achieves high-accuracy detection and correction of object shapes and positions, enabling effective collision avoidance and autonomous driving by predicting potential collisions and adjusting vehicle maneuvers accordingly.
Implementation Method 1
a second sensor that detects a distance to the object
Data Source
AI summary
An object recognition device includes a first sensor that acquires an image of an object present outside a movable body, a second sensor that detects a distance to the object, and a recognition unit that recognizes a position of the object with respect to the movable body and a shape of the object at the position by using outputs from the first sensor and the second sensor. If determining that the object is a first type object having a size equal to or more than a predetermined size, the recognition unit determines an actual shape of the object based on an image of the object acquired by the first sensor and the distance detected by the second sensor, and recognizes that the object is present as the actual shape at a position distanced from the movable body by the distance.